skills

Add tags and tags_match parameters to recall/reflect methods for
filtering
memories by visibility scope. Also add tags support to retain methods.

Changes:
- recall()/arecall(): add tags, tags_match parameters
- reflect()/areflect(): add tags, tags_match parameters
- retain()/aretain(): add tags parameter
- retain_batch()/aretain_batch(): add document_tags parameter
- Add TestTags test class with 7 tests

helm/hindsight/templates/_helpers.tpl

@@ -80,6 +80,22 @@ Control plane selector labels
 app.kubernetes.io/component: control-plane
 {{- end }}
 
+{{/*
+Worker labels
+*/}}
+{{- define "hindsight.worker.labels" -}}
+{{ include "hindsight.labels" . }}
+app.kubernetes.io/component: worker
+{{- end }}
+
+{{/*
+Worker selector labels
+*/}}
+{{- define "hindsight.worker.selectorLabels" -}}
+{{ include "hindsight.selectorLabels" . }}
+app.kubernetes.io/component: worker
+{{- end }}
+
 {{/*
 Create the name of the service account to use
 */}}

helm/hindsight/templates/api-deployment.yaml

@@ -55,6 +55,11 @@ spec:
         {{- end }}
         - name: HINDSIGHT_API_DATABASE_URL
           value: {{ include "hindsight.databaseUrl" . | quote }}
+        {{- /* Disable internal worker when dedicated workers are enabled */}}
+        {{- if .Values.worker.enabled }}
+        - name: HINDSIGHT_API_WORKER_ENABLED
+          value: "false"
+        {{- end }}
         {{- range $key, $value := .Values.api.env }}
         - name: {{ $key }}
           value: {{ $value | quote }}

helm/hindsight/templates/worker-service.yaml

@@ -0,0 +1,25 @@
+{{- if .Values.worker.enabled }}
+apiVersion: v1
+kind: Service
+metadata:
+  name: {{ include "hindsight.fullname" . }}-worker
+  labels:
+    {{- include "hindsight.worker.labels" . | nindent 4 }}
+  {{- if .Values.podAnnotations }}
+  annotations:
+    {{- /* Common Prometheus annotations for metrics scraping */}}
+    prometheus.io/scrape: "true"
+    prometheus.io/port: {{ .Values.worker.service.port | quote }}
+    prometheus.io/path: "/metrics"
+  {{- end }}
+spec:
+  # Headless service for StatefulSet (enables stable DNS names like worker-0.worker.namespace)
+  clusterIP: None
+  ports:
+  - port: {{ .Values.worker.service.port }}
+    targetPort: {{ .Values.worker.service.targetPort }}
+    protocol: TCP
+    name: http
+  selector:
+    {{- include "hindsight.worker.selectorLabels" . | nindent 4 }}
+{{- end }}

helm/hindsight/templates/worker-statefulset.yaml

@@ -0,0 +1,110 @@
+{{- if .Values.worker.enabled }}
+apiVersion: apps/v1
+kind: StatefulSet
+metadata:
+  name: {{ include "hindsight.fullname" . }}-worker
+  labels:
+    {{- include "hindsight.worker.labels" . | nindent 4 }}
+spec:
+  serviceName: {{ include "hindsight.fullname" . }}-worker
+  replicas: {{ .Values.worker.replicaCount }}
+  selector:
+    matchLabels:
+      {{- include "hindsight.worker.selectorLabels" . | nindent 6 }}
+  template:
+    metadata:
+      annotations:
+        {{- if not .Values.existingSecret }}
+        checksum/secret: {{ include (print $.Template.BasePath "/secret.yaml") . | sha256sum }}
+        {{- end }}
+        {{- with .Values.podAnnotations }}
+        {{- toYaml . | nindent 8 }}
+        {{- end }}
+      labels:
+        {{- include "hindsight.worker.selectorLabels" . | nindent 8 }}
+    spec:
+      {{- if .Values.serviceAccount.create }}
+      serviceAccountName: {{ include "hindsight.serviceAccountName" . }}
+      {{- end }}
+      securityContext:
+        {{- toYaml .Values.podSecurityContext | nindent 8 }}
+      containers:
+      - name: worker
+        securityContext:
+          {{- toYaml .Values.securityContext | nindent 10 }}
+        image: "{{ .Values.worker.image.repository }}:{{ .Values.worker.image.tag | default .Values.version }}"
+        imagePullPolicy: {{ .Values.worker.image.pullPolicy }}
+        command: ["hindsight-worker"]
+        ports:
+        - name: http
+          containerPort: {{ .Values.worker.service.targetPort }}
+          protocol: TCP
+        {{- if .Values.existingSecret }}
+        envFrom:
+        - secretRef:
+            name: {{ .Values.existingSecret }}
+        {{- end }}
+        env:
+        {{- /* POSTGRES_PASSWORD must be defined before DATABASE_URL for $(VAR) interpolation */}}
+        {{- if not .Values.postgresql.enabled }}
+        - name: POSTGRES_PASSWORD
+          valueFrom:
+            secretKeyRef:
+              name: {{ include "hindsight.secretName" . }}
+              key: postgres-password
+        {{- end }}
+        - name: HINDSIGHT_API_DATABASE_URL
+          value: {{ include "hindsight.databaseUrl" . | quote }}
+        {{- /* Worker ID uses pod name (StatefulSet provides stable names like worker-0, worker-1) */}}
+        - name: HINDSIGHT_API_WORKER_ID
+          valueFrom:
+            fieldRef:
+              fieldPath: metadata.name
+        {{- /* Inherit LLM config from api.env */}}
+        {{- range $key, $value := .Values.api.env }}
+        - name: {{ $key }}
+          value: {{ $value | quote }}
+        {{- end }}
+        {{- /* Worker-specific env vars */}}
+        {{- range $key, $value := .Values.worker.env }}
+        - name: {{ $key }}
+          value: {{ $value | quote }}
+        {{- end }}
+        {{- /* Only use secrets when not using existingSecret */}}
+        {{- if not .Values.existingSecret }}
+        {{- /* Inherit secrets from api.secrets */}}
+        {{- range $key, $value := .Values.api.secrets }}
+        - name: {{ $key }}
+          valueFrom:
+            secretKeyRef:
+              name: {{ include "hindsight.secretName" $ }}
+              key: {{ $key }}
+        {{- end }}
+        {{- /* Worker-specific secrets (can override api.secrets) */}}
+        {{- range $key, $value := .Values.worker.secrets }}
+        - name: {{ $key }}
+          valueFrom:
+            secretKeyRef:
+              name: {{ include "hindsight.secretName" $ }}
+              key: {{ $key }}
+        {{- end }}
+        {{- end }}
+        livenessProbe:
+          {{- toYaml .Values.worker.livenessProbe | nindent 10 }}
+        readinessProbe:
+          {{- toYaml .Values.worker.readinessProbe | nindent 10 }}
+        resources:
+          {{- toYaml .Values.worker.resources | nindent 10 }}
+      {{- with .Values.nodeSelector }}
+      nodeSelector:
+        {{- toYaml . | nindent 8 }}
+      {{- end }}
+      {{- with .Values.affinity }}
+      affinity:
+        {{- toYaml . | nindent 8 }}
+      {{- end }}
+      {{- with .Values.tolerations }}
+      tolerations:
+        {{- toYaml . | nindent 8 }}
+      {{- end }}
+{{- end }}

helm/hindsight/values.yaml

@@ -67,6 +67,63 @@ api:
     # HINDSIGHT_API_LLM_API_KEY: "your-api-key"
     # HINDSIGHT_API_LLM_BASE_URL: "https://api.groq.com/openai/v1"
 
+# Worker settings (distributed task processing)
+# When enabled, dedicated worker pods process tasks and the API's internal worker is disabled
+worker:
+  enabled: false
+  replicaCount: 2
+  image:
+    repository: ghcr.io/vectorize-io/hindsight-api
+    pullPolicy: IfNotPresent
+    # tag defaults to .Values.version if not specified
+
+  service:
+    # Service for metrics scraping (headless for StatefulSet)
+    port: 8889
+    targetPort: 8889
+
+  # Resource limits and requests
+  resources:
+    limits:
+      cpu: 2000m
+      memory: 4Gi
+    requests:
+      cpu: 500m
+      memory: 1Gi
+
+  # Liveness and readiness probes
+  livenessProbe:
+    httpGet:
+      path: /health
+      port: 8889
+    initialDelaySeconds: 30
+    periodSeconds: 10
+    timeoutSeconds: 5
+    failureThreshold: 3
+
+  readinessProbe:
+    httpGet:
+      path: /health
+      port: 8889
+    initialDelaySeconds: 10
+    periodSeconds: 5
+    timeoutSeconds: 3
+    failureThreshold: 3
+
+  # Worker-specific environment variables
+  env:
+    # Poll interval in milliseconds (how often to check for new tasks)
+    HINDSIGHT_API_WORKER_POLL_INTERVAL_MS: "500"
+    # Number of tasks to claim per poll cycle
+    HINDSIGHT_API_WORKER_BATCH_SIZE: "10"
+    # Max retries before marking a task as failed
+    HINDSIGHT_API_WORKER_MAX_RETRIES: "3"
+    # HTTP port for metrics/health (matches service.targetPort)
+    HINDSIGHT_API_WORKER_HTTP_PORT: "8889"
+
+  # Secret environment variables (inherited from api.secrets if not specified)
+  secrets: {}
+
 # Image settings for control plane
 controlPlane:
   enabled: true

hindsight-api/hindsight_api/admin/cli.py

@@ -244,6 +244,65 @@ def run_db_migration(
     typer.echo("Database migrations completed successfully")
 
 
+async def _decommission_worker(db_url: str, worker_id: str, schema: str = "public") -> int:
+    """Release all tasks owned by a worker, setting them back to pending status."""
+    is_pg0, instance_name, _ = parse_pg0_url(db_url)
+    if is_pg0:
+        typer.echo(f"Starting embedded PostgreSQL (instance: {instance_name})...")
+    resolved_url = await resolve_database_url(db_url)
+
+    conn = await asyncpg.connect(resolved_url)
+    try:
+        table = _fq_table("async_operations", schema)
+        result = await conn.fetch(
+            f"""
+            UPDATE {table}
+            SET status = 'pending', worker_id = NULL, claimed_at = NULL, updated_at = now()
+            WHERE worker_id = $1 AND status = 'processing'
+            RETURNING operation_id
+            """,
+            worker_id,
+        )
+        return len(result)
+    finally:
+        await conn.close()
+
+
+@app.command(name="decommission-worker")
+def decommission_worker(
+    worker_id: str = typer.Argument(..., help="Worker ID to decommission"),
+    schema: str = typer.Option("public", "--schema", "-s", help="Database schema"),
+    yes: bool = typer.Option(False, "--yes", "-y", help="Skip confirmation prompt"),
+):
+    """Release all tasks owned by a worker (sets status back to pending).
+
+    Use this command when a worker has crashed or been removed without graceful shutdown.
+    All tasks that were being processed by the worker will be released back to the queue
+    so other workers can pick them up.
+    """
+    config = HindsightConfig.from_env()
+
+    if not config.database_url:
+        typer.echo("Error: Database URL not configured.", err=True)
+        typer.echo("Set HINDSIGHT_API_DATABASE_URL environment variable.", err=True)
+        raise typer.Exit(1)
+
+    if not yes:
+        typer.confirm(
+            f"This will release all tasks owned by worker '{worker_id}' back to pending. Continue?",
+            abort=True,
+        )
+
+    typer.echo(f"Decommissioning worker '{worker_id}' (schema: {schema})...")
+
+    count = asyncio.run(_decommission_worker(config.database_url, worker_id, schema))
+
+    if count > 0:
+        typer.echo(f"Released {count} task(s) from worker '{worker_id}'")
+    else:
+        typer.echo(f"No tasks found for worker '{worker_id}'")
+
+
 def main():
     app()
 

hindsight-api/hindsight_api/alembic/versions/l7g8h9i0j1k2_add_worker_columns.py

@@ -0,0 +1,109 @@
+"""add_worker_columns
+
+Revision ID: l7g8h9i0j1k2
+Revises: k6f7g8h9i0j1
+Create Date: 2026-01-19 00:00:00.000000
+
+This migration adds columns to async_operations for distributed worker support:
+- worker_id: ID of the worker that claimed the task
+- claimed_at: When the task was claimed
+- retry_count: Number of retry attempts
+- task_payload: The serialized task dictionary
+"""
+
+from collections.abc import Sequence
+
+import sqlalchemy as sa
+from alembic import context, op
+from sqlalchemy.dialects import postgresql
+
+# revision identifiers, used by Alembic.
+revision: str = "l7g8h9i0j1k2"
+down_revision: str | Sequence[str] | None = "k6f7g8h9i0j1"
+branch_labels: str | Sequence[str] | None = None
+depends_on: str | Sequence[str] | None = None
+
+
+def _get_schema_prefix() -> str:
+    """Get schema prefix for table names (required for multi-tenant support)."""
+    schema = context.config.get_main_option("target_schema")
+    return f'"{schema}".' if schema else ""
+
+
+def upgrade() -> None:
+    """Add worker columns to async_operations."""
+    schema = _get_schema_prefix()
+
+    # Add worker_id column (ID of worker that claimed the task)
+    op.add_column(
+        "async_operations",
+        sa.Column("worker_id", sa.Text(), nullable=True),
+        schema=context.config.get_main_option("target_schema") or None,
+    )
+
+    # Add claimed_at column (when task was claimed by worker)
+    op.add_column(
+        "async_operations",
+        sa.Column("claimed_at", postgresql.TIMESTAMP(timezone=True), nullable=True),
+        schema=context.config.get_main_option("target_schema") or None,
+    )
+
+    # Add retry_count column (number of retry attempts)
+    op.add_column(
+        "async_operations",
+        sa.Column("retry_count", sa.Integer(), server_default="0", nullable=False),
+        schema=context.config.get_main_option("target_schema") or None,
+    )
+
+    # Add task_payload column (serialized task dictionary)
+    op.add_column(
+        "async_operations",
+        sa.Column(
+            "task_payload",
+            postgresql.JSONB(astext_type=sa.Text()),
+            nullable=True,
+        ),
+        schema=context.config.get_main_option("target_schema") or None,
+    )
+
+    # Add index for efficient worker polling (pending tasks ordered by creation time)
+    op.execute(
+        f"CREATE INDEX idx_async_operations_pending_claim ON {schema}async_operations (status, created_at) "
+        f"WHERE status = 'pending' AND task_payload IS NOT NULL"
+    )
+
+    # Add index for finding tasks by worker_id (for decommissioning)
+    op.execute(
+        f"CREATE INDEX idx_async_operations_worker_id ON {schema}async_operations (worker_id) WHERE worker_id IS NOT NULL"
+    )
+
+
+def downgrade() -> None:
+    """Remove worker columns from async_operations."""
+    schema = _get_schema_prefix()
+
+    # Drop indexes
+    op.execute(f"DROP INDEX IF EXISTS {schema}idx_async_operations_pending_claim")
+    op.execute(f"DROP INDEX IF EXISTS {schema}idx_async_operations_worker_id")
+
+    # Drop columns
+    op.drop_column(
+        "async_operations",
+        "task_payload",
+        schema=context.config.get_main_option("target_schema") or None,
+    )
+    op.drop_column(
+        "async_operations",
+        "retry_count",
+        schema=context.config.get_main_option("target_schema") or None,
+    )
+    op.drop_column(
+        "async_operations",
+        "claimed_at",
+        schema=context.config.get_main_option("target_schema") or None,
+    )
+    op.drop_column(
+        "async_operations",
+        "worker_id",
+        schema=context.config.get_main_option("target_schema") or None,
+    )

hindsight-api/hindsight_api/alembic/versions/m8h9i0j1k2l3_mental_model_id_to_text.py

@@ -0,0 +1,41 @@
+"""mental_model_id_to_text
+
+Revision ID: m8h9i0j1k2l3
+Revises: l7g8h9i0j1k2
+Create Date: 2026-01-19 00:00:00.000000
+
+This migration changes the mental_models.id column from VARCHAR(64) to TEXT
+to support longer model IDs (e.g., entity names that exceed 64 characters).
+"""
+
+from collections.abc import Sequence
+
+from alembic import context, op
+
+# revision identifiers, used by Alembic.
+revision: str = "m8h9i0j1k2l3"
+down_revision: str | Sequence[str] | None = "l7g8h9i0j1k2"
+branch_labels: str | Sequence[str] | None = None
+depends_on: str | Sequence[str] | None = None
+
+
+def _get_schema_prefix() -> str:
+    """Get schema prefix for table names (required for multi-tenant support)."""
+    schema = context.config.get_main_option("target_schema")
+    return f'"{schema}".' if schema else ""
+
+
+def upgrade() -> None:
+    """Change mental_models.id from VARCHAR(64) to TEXT."""
+    schema = _get_schema_prefix()
+
+    # Alter the id column type from VARCHAR(64) to TEXT
+    op.execute(f"ALTER TABLE {schema}mental_models ALTER COLUMN id TYPE TEXT")
+
+
+def downgrade() -> None:
+    """Revert mental_models.id from TEXT to VARCHAR(64)."""
+    schema = _get_schema_prefix()
+
+    # Note: This may fail if any id values exceed 64 characters
+    op.execute(f"ALTER TABLE {schema}mental_models ALTER COLUMN id TYPE VARCHAR(64)")

hindsight-api/hindsight_api/alembic/versions/n9i0j1k2l3m4_learnings_and_pinned_reflections.py

@@ -0,0 +1,134 @@
+"""learnings_and_pinned_reflections
+
+Revision ID: n9i0j1k2l3m4
+Revises: m8h9i0j1k2l3
+Create Date: 2026-01-21 00:00:00.000000
+
+This migration:
+1. Creates the 'learnings' table for automatic bottom-up consolidation
+2. Creates the 'pinned_reflections' table for user-curated living documents
+3. Adds consolidation tracking columns to the 'banks' table
+"""
+
+from collections.abc import Sequence
+
+from alembic import context, op
+
+# revision identifiers, used by Alembic.
+revision: str = "n9i0j1k2l3m4"
+down_revision: str | Sequence[str] | None = "m8h9i0j1k2l3"
+branch_labels: str | Sequence[str] | None = None
+depends_on: str | Sequence[str] | None = None
+
+
+def _get_schema_prefix() -> str:
+    """Get schema prefix for table names (required for multi-tenant support)."""
+    schema = context.config.get_main_option("target_schema")
+    return f'"{schema}".' if schema else ""
+
+
+def upgrade() -> None:
+    """Create learnings and pinned_reflections tables."""
+    schema = _get_schema_prefix()
+
+    # 1. Create learnings table
+    op.execute(f"""
+        CREATE TABLE {schema}learnings (
+            id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
+            bank_id VARCHAR(64) NOT NULL,
+            text TEXT NOT NULL,
+            proof_count INT NOT NULL DEFAULT 1,
+            history JSONB DEFAULT '[]'::jsonb,
+            mission_context VARCHAR(64),
+            pre_mission_change BOOLEAN DEFAULT FALSE,
+            embedding vector(384),
+            tags VARCHAR[] DEFAULT ARRAY[]::VARCHAR[],
+            created_at TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT now(),
+            updated_at TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT now()
+        )
+    """)
+
+    # Add foreign key constraint
+    op.execute(f"""
+        ALTER TABLE {schema}learnings
+        ADD CONSTRAINT fk_learnings_bank_id
+        FOREIGN KEY (bank_id) REFERENCES {schema}banks(bank_id) ON DELETE CASCADE
+    """)
+
+    # Indexes for learnings
+    op.execute(f"CREATE INDEX idx_learnings_bank_id ON {schema}learnings(bank_id)")
+    op.execute(f"""
+        CREATE INDEX idx_learnings_embedding ON {schema}learnings
+        USING hnsw (embedding vector_cosine_ops)
+    """)
+    op.execute(f"CREATE INDEX idx_learnings_tags ON {schema}learnings USING GIN(tags)")
+
+    # Full-text search for learnings
+    op.execute(f"""
+        ALTER TABLE {schema}learnings ADD COLUMN search_vector tsvector
+        GENERATED ALWAYS AS (to_tsvector('english', text)) STORED
+    """)
+    op.execute(f"CREATE INDEX idx_learnings_text_search ON {schema}learnings USING gin(search_vector)")
+
+    # 2. Create pinned_reflections table
+    op.execute(f"""
+        CREATE TABLE {schema}pinned_reflections (
+            id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
+            bank_id VARCHAR(64) NOT NULL,
+            name VARCHAR(256) NOT NULL,
+            source_query TEXT NOT NULL,
+            content TEXT NOT NULL,
+            embedding vector(384),
+            tags VARCHAR[] DEFAULT ARRAY[]::VARCHAR[],
+            last_refreshed_at TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT now(),
+            created_at TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT now()
+        )
+    """)
+
+    # Add foreign key constraint
+    op.execute(f"""
+        ALTER TABLE {schema}pinned_reflections
+        ADD CONSTRAINT fk_pinned_reflections_bank_id
+        FOREIGN KEY (bank_id) REFERENCES {schema}banks(bank_id) ON DELETE CASCADE
+    """)
+
+    # Indexes for pinned_reflections
+    op.execute(f"CREATE INDEX idx_pinned_reflections_bank_id ON {schema}pinned_reflections(bank_id)")
+    op.execute(f"""
+        CREATE INDEX idx_pinned_reflections_embedding ON {schema}pinned_reflections
+        USING hnsw (embedding vector_cosine_ops)
+    """)
+    op.execute(f"CREATE INDEX idx_pinned_reflections_tags ON {schema}pinned_reflections USING GIN(tags)")
+
+    # Full-text search for pinned_reflections
+    op.execute(f"""
+        ALTER TABLE {schema}pinned_reflections ADD COLUMN search_vector tsvector
+        GENERATED ALWAYS AS (to_tsvector('english', COALESCE(name, '') || ' ' || content)) STORED
+    """)
+    op.execute(f"""
+        CREATE INDEX idx_pinned_reflections_text_search ON {schema}pinned_reflections
+        USING gin(search_vector)
+    """)
+
+    # 3. Add consolidation tracking columns to banks table
+    op.execute(f"""
+        ALTER TABLE {schema}banks
+        ADD COLUMN IF NOT EXISTS last_consolidated_at TIMESTAMP WITH TIME ZONE
+    """)
+    op.execute(f"""
+        ALTER TABLE {schema}banks
+        ADD COLUMN IF NOT EXISTS mission_changed_at TIMESTAMP WITH TIME ZONE
+    """)
+
+
+def downgrade() -> None:
+    """Drop learnings and pinned_reflections tables."""
+    schema = _get_schema_prefix()
+
+    # Drop tables
+    op.execute(f"DROP TABLE IF EXISTS {schema}learnings CASCADE")
+    op.execute(f"DROP TABLE IF EXISTS {schema}pinned_reflections CASCADE")
+
+    # Remove columns from banks
+    op.execute(f"ALTER TABLE {schema}banks DROP COLUMN IF EXISTS last_consolidated_at")
+    op.execute(f"ALTER TABLE {schema}banks DROP COLUMN IF EXISTS mission_changed_at")

hindsight-api/hindsight_api/alembic/versions/o0j1k2l3m4n5_migrate_mental_models_data.py

@@ -0,0 +1,113 @@
+"""migrate_mental_models_data
+
+Revision ID: o0j1k2l3m4n5
+Revises: n9i0j1k2l3m4
+Create Date: 2026-01-21 00:00:00.000000
+
+This migration:
+1. Migrates existing 'pinned' mental models to the new 'pinned_reflections' table
+2. Migrates existing 'learned' mental models to the new 'learnings' table
+3. Deletes non-directive mental models (structural, emergent, pinned, learned)
+4. Drops the mental_model_versions table (no longer used)
+5. Adds a CHECK constraint that only 'directive' subtype is allowed
+"""
+
+from collections.abc import Sequence
+
+from alembic import context, op
+
+# revision identifiers, used by Alembic.
+revision: str = "o0j1k2l3m4n5"
+down_revision: str | Sequence[str] | None = "n9i0j1k2l3m4"
+branch_labels: str | Sequence[str] | None = None
+depends_on: str | Sequence[str] | None = None
+
+
+def _get_schema_prefix() -> str:
+    """Get schema prefix for table names (required for multi-tenant support)."""
+    schema = context.config.get_main_option("target_schema")
+    return f'"{schema}".' if schema else ""
+
+
+def upgrade() -> None:
+    """Migrate data and clean up old mental models."""
+    schema = _get_schema_prefix()
+
+    # 1. Migrate 'pinned' mental models to pinned_reflections
+    # For pinned models, the first observation's content becomes the pinned reflection content
+    op.execute(f"""
+        INSERT INTO {schema}pinned_reflections (bank_id, name, source_query, content, tags, created_at)
+        SELECT
+            bank_id,
+            name,
+            description AS source_query,
+            COALESCE(
+                observations->'observations'->0->>'content',
+                description,
+                ''
+            ) AS content,
+            tags,
+            created_at
+        FROM {schema}mental_models
+        WHERE subtype = 'pinned'
+        ON CONFLICT DO NOTHING
+    """)
+
+    # 2. Migrate 'learned' mental models to learnings
+    # Each observation in a learned model becomes a separate learning
+    op.execute(f"""
+        INSERT INTO {schema}learnings (bank_id, text, proof_count, tags, created_at)
+        SELECT
+            mm.bank_id,
+            obs->>'content' AS text,
+            GREATEST(1, COALESCE(jsonb_array_length(obs->'evidence'), 1)) AS proof_count,
+            mm.tags,
+            mm.created_at
+        FROM {schema}mental_models mm,
+        LATERAL jsonb_array_elements(mm.observations->'observations') AS obs
+        WHERE mm.subtype = 'learned'
+          AND obs->>'content' IS NOT NULL
+          AND obs->>'content' != ''
+        ON CONFLICT DO NOTHING
+    """)
+
+    # 3. Delete all non-directive mental models (they've been migrated or are obsolete)
+    op.execute(f"""
+        DELETE FROM {schema}mental_models
+        WHERE subtype != 'directive'
+    """)
+
+    # 4. Drop the mental_model_versions table (no longer used)
+    op.execute(f"DROP TABLE IF EXISTS {schema}mental_model_versions CASCADE")
+
+    # 5. Drop old constraints and add new one that only allows 'directive'
+    op.execute(f"ALTER TABLE {schema}mental_models DROP CONSTRAINT IF EXISTS ck_mental_models_subtype")
+    op.execute(f"""
+        ALTER TABLE {schema}mental_models
+        ADD CONSTRAINT ck_mental_models_subtype CHECK (subtype = 'directive')
+    """)
+
+
+def downgrade() -> None:
+    """Reverse the migration (data migration is one-way, so this just removes constraints)."""
+    schema = _get_schema_prefix()
+
+    # Remove the directive-only constraint
+    op.execute(f"ALTER TABLE {schema}mental_models DROP CONSTRAINT IF EXISTS ck_mental_models_subtype")
+
+    # Re-create mental_model_versions table
+    op.execute(f"""
+        CREATE TABLE IF NOT EXISTS {schema}mental_model_versions (
+            id SERIAL PRIMARY KEY,
+            bank_id VARCHAR(64) NOT NULL,
+            model_id VARCHAR(128) NOT NULL,
+            version INT NOT NULL,
+            observations JSONB NOT NULL,
+            created_at TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT now()
+        )
+    """)
+    op.execute(
+        f"CREATE INDEX IF NOT EXISTS idx_mm_versions_lookup ON {schema}mental_model_versions(bank_id, model_id, version DESC)"
+    )
+
+    # Note: Data migration cannot be reversed - pinned_reflections and learnings data remains

hindsight-api/hindsight_api/alembic/versions/p1k2l3m4n5o6_new_knowledge_architecture.py

@@ -0,0 +1,194 @@
+"""new_knowledge_architecture
+
+Revision ID: p1k2l3m4n5o6
+Revises: o0j1k2l3m4n5
+Create Date: 2026-01-21 00:00:00.000000
+
+This migration implements the new knowledge architecture:
+1. Drops the 'learnings' table (mental models are now in memory_units)
+2. Renames 'pinned_reflections' to 'reflections'
+3. Drops the 'mental_models' table completely
+4. Creates 'directives' table for hard rules
+5. Adds mental model support columns to 'memory_units' (proof_count, source_memory_ids, history)
+
+The new architecture:
+- Directives: Hard rules in their own table
+- Mental Models: Stored in memory_units with fact_type='mental_model'
+- Reflections: User-curated documents (renamed from pinned_reflections)
+"""
+
+from collections.abc import Sequence
+
+from alembic import context, op
+
+# revision identifiers, used by Alembic.
+revision: str = "p1k2l3m4n5o6"
+down_revision: str | Sequence[str] | None = "o0j1k2l3m4n5"
+branch_labels: str | Sequence[str] | None = None
+depends_on: str | Sequence[str] | None = None
+
+
+def _get_schema_prefix() -> str:
+    """Get schema prefix for table names (required for multi-tenant support)."""
+    schema = context.config.get_main_option("target_schema")
+    return f'"{schema}".' if schema else ""
+
+
+def upgrade() -> None:
+    """Implement new knowledge architecture."""
+    schema = _get_schema_prefix()
+
+    # 1. Drop the learnings table (mental models will be in memory_units)
+    op.execute(f"DROP TABLE IF EXISTS {schema}learnings CASCADE")
+
+    # 2. Rename pinned_reflections to reflections
+    op.execute(f"ALTER TABLE IF EXISTS {schema}pinned_reflections RENAME TO reflections")
+
+    # Rename indexes for reflections
+    op.execute(f"ALTER INDEX IF EXISTS {schema}idx_pinned_reflections_bank_id RENAME TO idx_reflections_bank_id")
+    op.execute(f"ALTER INDEX IF EXISTS {schema}idx_pinned_reflections_embedding RENAME TO idx_reflections_embedding")
+    op.execute(f"ALTER INDEX IF EXISTS {schema}idx_pinned_reflections_tags RENAME TO idx_reflections_tags")
+    op.execute(
+        f"ALTER INDEX IF EXISTS {schema}idx_pinned_reflections_text_search RENAME TO idx_reflections_text_search"
+    )
+
+    # Rename foreign key constraint
+    op.execute(f"""
+        ALTER TABLE {schema}reflections
+        DROP CONSTRAINT IF EXISTS fk_pinned_reflections_bank_id
+    """)
+    op.execute(f"""
+        ALTER TABLE {schema}reflections
+        ADD CONSTRAINT fk_reflections_bank_id
+        FOREIGN KEY (bank_id) REFERENCES {schema}banks(bank_id) ON DELETE CASCADE
+    """)
+
+    # 3. Drop the mental_models table completely
+    op.execute(f"DROP TABLE IF EXISTS {schema}mental_models CASCADE")
+
+    # 4. Create directives table
+    op.execute(f"""
+        CREATE TABLE {schema}directives (
+            id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
+            bank_id VARCHAR(64) NOT NULL,
+            name VARCHAR(256) NOT NULL,
+            content TEXT NOT NULL,
+            priority INT NOT NULL DEFAULT 0,
+            is_active BOOLEAN NOT NULL DEFAULT TRUE,
+            tags VARCHAR[] DEFAULT ARRAY[]::VARCHAR[],
+            created_at TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT now(),
+            updated_at TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT now()
+        )
+    """)
+
+    # Add foreign key and indexes for directives
+    op.execute(f"""
+        ALTER TABLE {schema}directives
+        ADD CONSTRAINT fk_directives_bank_id
+        FOREIGN KEY (bank_id) REFERENCES {schema}banks(bank_id) ON DELETE CASCADE
+    """)
+    op.execute(f"CREATE INDEX idx_directives_bank_id ON {schema}directives(bank_id)")
+    op.execute(f"CREATE INDEX idx_directives_bank_active ON {schema}directives(bank_id, is_active)")
+    op.execute(f"CREATE INDEX idx_directives_tags ON {schema}directives USING GIN(tags)")
+
+    # 5. Add mental model support columns to memory_units
+    # proof_count: Number of memories that support this mental model
+    op.execute(f"""
+        ALTER TABLE {schema}memory_units
+        ADD COLUMN IF NOT EXISTS proof_count INT DEFAULT 1
+    """)
+
+    # source_memory_ids: Array of memory IDs that consolidated into this mental model
+    op.execute(f"""
+        ALTER TABLE {schema}memory_units
+        ADD COLUMN IF NOT EXISTS source_memory_ids UUID[] DEFAULT ARRAY[]::UUID[]
+    """)
+
+    # history: JSONB array tracking changes to mental models
+    op.execute(f"""
+        ALTER TABLE {schema}memory_units
+        ADD COLUMN IF NOT EXISTS history JSONB DEFAULT '[]'::jsonb
+    """)
+
+    # Add index for finding mental models
+    op.execute(f"""
+        CREATE INDEX IF NOT EXISTS idx_memory_units_mental_models
+        ON {schema}memory_units(bank_id, fact_type)
+        WHERE fact_type = 'mental_model'
+    """)
+
+    # 6. Update fact_type check constraint to include 'mental_model'
+    op.execute(f"ALTER TABLE {schema}memory_units DROP CONSTRAINT IF EXISTS memory_units_fact_type_check")
+    op.execute(f"""
+        ALTER TABLE {schema}memory_units
+        ADD CONSTRAINT memory_units_fact_type_check
+        CHECK (fact_type IN ('world', 'experience', 'opinion', 'observation', 'mental_model'))
+    """)
+
+
+def downgrade() -> None:
+    """Reverse the migration."""
+    schema = _get_schema_prefix()
+
+    # Restore original fact_type check constraint (without 'mental_model')
+    op.execute(f"ALTER TABLE {schema}memory_units DROP CONSTRAINT IF EXISTS memory_units_fact_type_check")
+    op.execute(f"""
+        ALTER TABLE {schema}memory_units
+        ADD CONSTRAINT memory_units_fact_type_check
+        CHECK (fact_type IN ('world', 'experience', 'opinion', 'observation'))
+    """)
+
+    # Drop mental model columns from memory_units
+    op.execute(f"ALTER TABLE {schema}memory_units DROP COLUMN IF EXISTS proof_count")
+    op.execute(f"ALTER TABLE {schema}memory_units DROP COLUMN IF EXISTS source_memory_ids")
+    op.execute(f"ALTER TABLE {schema}memory_units DROP COLUMN IF EXISTS history")
+    op.execute(f"DROP INDEX IF EXISTS {schema}idx_memory_units_mental_models")
+
+    # Drop directives table
+    op.execute(f"DROP TABLE IF EXISTS {schema}directives CASCADE")
+
+    # Rename reflections back to pinned_reflections
+    op.execute(f"ALTER TABLE IF EXISTS {schema}reflections RENAME TO pinned_reflections")
+
+    # Restore indexes
+    op.execute(f"ALTER INDEX IF EXISTS {schema}idx_reflections_bank_id RENAME TO idx_pinned_reflections_bank_id")
+    op.execute(f"ALTER INDEX IF EXISTS {schema}idx_reflections_embedding RENAME TO idx_pinned_reflections_embedding")
+    op.execute(f"ALTER INDEX IF EXISTS {schema}idx_reflections_tags RENAME TO idx_pinned_reflections_tags")
+    op.execute(
+        f"ALTER INDEX IF EXISTS {schema}idx_reflections_text_search RENAME TO idx_pinned_reflections_text_search"
+    )
+
+    # Restore foreign key
+    op.execute(f"""
+        ALTER TABLE {schema}pinned_reflections
+        DROP CONSTRAINT IF EXISTS fk_reflections_bank_id
+    """)
+    op.execute(f"""
+        ALTER TABLE {schema}pinned_reflections
+        ADD CONSTRAINT fk_pinned_reflections_bank_id
+        FOREIGN KEY (bank_id) REFERENCES {schema}banks(bank_id) ON DELETE CASCADE
+    """)
+
+    # Re-create learnings table
+    op.execute(f"""
+        CREATE TABLE IF NOT EXISTS {schema}learnings (
+            id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
+            bank_id VARCHAR(64) NOT NULL,
+            text TEXT NOT NULL,
+            proof_count INT NOT NULL DEFAULT 1,
+            history JSONB DEFAULT '[]'::jsonb,
+            mission_context VARCHAR(64),
+            pre_mission_change BOOLEAN DEFAULT FALSE,
+            embedding vector(384),
+            tags VARCHAR[] DEFAULT ARRAY[]::VARCHAR[],
+            created_at TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT now(),
+            updated_at TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT now()
+        )
+    """)
+    op.execute(f"""
+        ALTER TABLE {schema}learnings
+        ADD CONSTRAINT fk_learnings_bank_id
+        FOREIGN KEY (bank_id) REFERENCES {schema}banks(bank_id) ON DELETE CASCADE
+    """)
+
+    # Note: mental_models table recreation is complex and would need separate handling

hindsight-api/hindsight_api/alembic/versions/q2l3m4n5o6p7_fix_mental_model_fact_type.py

@@ -0,0 +1,50 @@
+"""fix_mental_model_fact_type
+
+Revision ID: q2l3m4n5o6p7
+Revises: p1k2l3m4n5o6
+Create Date: 2026-01-21 13:30:00.000000
+
+Fix the fact_type check constraint to include 'mental_model'.
+This is a fix for p1k2l3m4n5o6 which should have included this change.
+"""
+
+from collections.abc import Sequence
+
+from alembic import context, op
+
+# revision identifiers, used by Alembic.
+revision: str = "q2l3m4n5o6p7"
+down_revision: str | Sequence[str] | None = "p1k2l3m4n5o6"
+branch_labels: str | Sequence[str] | None = None
+depends_on: str | Sequence[str] | None = None
+
+
+def _get_schema_prefix() -> str:
+    """Get schema prefix for table names (required for multi-tenant support)."""
+    schema = context.config.get_main_option("target_schema")
+    return f'"{schema}".' if schema else ""
+
+
+def upgrade() -> None:
+    """Add 'mental_model' to the fact_type check constraint."""
+    schema = _get_schema_prefix()
+
+    # Drop the old constraint and add the new one with mental_model included
+    op.execute(f"ALTER TABLE {schema}memory_units DROP CONSTRAINT IF EXISTS memory_units_fact_type_check")
+    op.execute(f"""
+        ALTER TABLE {schema}memory_units
+        ADD CONSTRAINT memory_units_fact_type_check
+        CHECK (fact_type IN ('world', 'experience', 'opinion', 'observation', 'mental_model'))
+    """)
+
+
+def downgrade() -> None:
+    """Remove 'mental_model' from the fact_type check constraint."""
+    schema = _get_schema_prefix()
+
+    op.execute(f"ALTER TABLE {schema}memory_units DROP CONSTRAINT IF EXISTS memory_units_fact_type_check")
+    op.execute(f"""
+        ALTER TABLE {schema}memory_units
+        ADD CONSTRAINT memory_units_fact_type_check
+        CHECK (fact_type IN ('world', 'experience', 'opinion', 'observation'))
+    """)

hindsight-api/hindsight_api/alembic/versions/r3m4n5o6p7q8_add_reflect_response_to_reflections.py

@@ -0,0 +1,47 @@
+"""Add reflect_response JSONB column to reflections
+
+Revision ID: r3m4n5o6p7q8
+Revises: q2l3m4n5o6p7
+Create Date: 2026-01-21
+
+This migration adds a reflect_response JSONB column to store the full
+reflect API response payload, including based_on facts and trace data.
+
+Note: Table was renamed from pinned_reflections to reflections in p1k2l3m4n5o6.
+"""
+
+from collections.abc import Sequence
+
+from alembic import context, op
+
+revision: str = "r3m4n5o6p7q8"
+down_revision: str | Sequence[str] | None = "q2l3m4n5o6p7"
+branch_labels: str | Sequence[str] | None = None
+depends_on: str | Sequence[str] | None = None
+
+
+def _get_schema_prefix() -> str:
+    """Get schema prefix for table names (required for multi-tenant support)."""
+    schema = context.config.get_main_option("target_schema")
+    return f'"{schema}".' if schema else ""
+
+
+def upgrade() -> None:
+    """Add reflect_response JSONB column to reflections."""
+    schema = _get_schema_prefix()
+
+    # Add reflect_response column to store the full reflect API response
+    op.execute(f"""
+        ALTER TABLE {schema}reflections
+        ADD COLUMN IF NOT EXISTS reflect_response JSONB
+    """)
+
+
+def downgrade() -> None:
+    """Remove reflect_response column from reflections."""
+    schema = _get_schema_prefix()
+
+    op.execute(f"""
+        ALTER TABLE {schema}reflections
+        DROP COLUMN IF EXISTS reflect_response
+    """)

hindsight-api/hindsight_api/config.py

@@ -93,6 +93,11 @@ ENV_RETAIN_EXTRACT_CAUSAL_LINKS = "HINDSIGHT_API_RETAIN_EXTRACT_CAUSAL_LINKS"
 ENV_RETAIN_EXTRACTION_MODE = "HINDSIGHT_API_RETAIN_EXTRACTION_MODE"
 ENV_RETAIN_OBSERVATIONS_ASYNC = "HINDSIGHT_API_RETAIN_OBSERVATIONS_ASYNC"
 
+# Mental models settings
+ENV_ENABLE_MENTAL_MODELS = "HINDSIGHT_API_ENABLE_MENTAL_MODELS"
+ENV_CONSOLIDATION_SIMILARITY_THRESHOLD = "HINDSIGHT_API_CONSOLIDATION_SIMILARITY_THRESHOLD"
+ENV_CONSOLIDATION_BATCH_SIZE = "HINDSIGHT_API_CONSOLIDATION_BATCH_SIZE"
+
 # Optimization flags
 ENV_SKIP_LLM_VERIFICATION = "HINDSIGHT_API_SKIP_LLM_VERIFICATION"
 ENV_LAZY_RERANKER = "HINDSIGHT_API_LAZY_RERANKER"
@@ -106,10 +111,13 @@ ENV_DB_POOL_MAX_SIZE = "HINDSIGHT_API_DB_POOL_MAX_SIZE"
 ENV_DB_COMMAND_TIMEOUT = "HINDSIGHT_API_DB_COMMAND_TIMEOUT"
 ENV_DB_ACQUIRE_TIMEOUT = "HINDSIGHT_API_DB_ACQUIRE_TIMEOUT"
 
-# Background task processing
-ENV_TASK_BACKEND = "HINDSIGHT_API_TASK_BACKEND"
-ENV_TASK_BACKEND_MEMORY_BATCH_SIZE = "HINDSIGHT_API_TASK_BACKEND_MEMORY_BATCH_SIZE"
-ENV_TASK_BACKEND_MEMORY_BATCH_INTERVAL = "HINDSIGHT_API_TASK_BACKEND_MEMORY_BATCH_INTERVAL"
+# Worker configuration (distributed task processing)
+ENV_WORKER_ENABLED = "HINDSIGHT_API_WORKER_ENABLED"
+ENV_WORKER_ID = "HINDSIGHT_API_WORKER_ID"
+ENV_WORKER_POLL_INTERVAL_MS = "HINDSIGHT_API_WORKER_POLL_INTERVAL_MS"
+ENV_WORKER_MAX_RETRIES = "HINDSIGHT_API_WORKER_MAX_RETRIES"
+ENV_WORKER_BATCH_SIZE = "HINDSIGHT_API_WORKER_BATCH_SIZE"
+ENV_WORKER_HTTP_PORT = "HINDSIGHT_API_WORKER_HTTP_PORT"
 
 # Reflect agent settings
 ENV_REFLECT_MAX_ITERATIONS = "HINDSIGHT_API_REFLECT_MAX_ITERATIONS"
@@ -168,6 +176,11 @@ DEFAULT_RETAIN_EXTRACTION_MODE = "concise"  # Extraction mode: "concise" or "ver
 RETAIN_EXTRACTION_MODES = ("concise", "verbose")  # Allowed extraction modes
 DEFAULT_RETAIN_OBSERVATIONS_ASYNC = False  # Run observation generation async (after retain completes)
 
+# Mental models defaults
+DEFAULT_ENABLE_MENTAL_MODELS = False  # Mental models disabled by default (experimental)
+DEFAULT_CONSOLIDATION_SIMILARITY_THRESHOLD = 0.75  # Minimum similarity to consider a learning related
+DEFAULT_CONSOLIDATION_BATCH_SIZE = 50  # Memories to load per batch (internal memory optimization)
+
 # Database migrations
 DEFAULT_RUN_MIGRATIONS_ON_STARTUP = True
 
@@ -177,10 +190,13 @@ DEFAULT_DB_POOL_MAX_SIZE = 100
 DEFAULT_DB_COMMAND_TIMEOUT = 60  # seconds
 DEFAULT_DB_ACQUIRE_TIMEOUT = 30  # seconds
 
-# Background task processing
-DEFAULT_TASK_BACKEND = "memory"  # Options: "memory", "noop"
-DEFAULT_TASK_BACKEND_MEMORY_BATCH_SIZE = 10
-DEFAULT_TASK_BACKEND_MEMORY_BATCH_INTERVAL = 1.0  # seconds
+# Worker configuration (distributed task processing)
+DEFAULT_WORKER_ENABLED = True  # API runs worker by default (standalone mode)
+DEFAULT_WORKER_ID = None  # Will use hostname if not specified
+DEFAULT_WORKER_POLL_INTERVAL_MS = 500  # Poll database every 500ms
+DEFAULT_WORKER_MAX_RETRIES = 3  # Max retries before marking task failed
+DEFAULT_WORKER_BATCH_SIZE = 10  # Tasks to claim per poll cycle
+DEFAULT_WORKER_HTTP_PORT = 8889  # HTTP port for worker metrics/health
 
 # Reflect agent settings
 DEFAULT_REFLECT_MAX_ITERATIONS = 10  # Max tool call iterations before forcing response
@@ -318,6 +334,11 @@ class HindsightConfig:
     retain_extraction_mode: str
     retain_observations_async: bool
 
+    # Mental models settings
+    enable_mental_models: bool
+    consolidation_similarity_threshold: float
+    consolidation_batch_size: int
+
     # Optimization flags
     skip_llm_verification: bool
     lazy_reranker: bool
@@ -331,10 +352,13 @@ class HindsightConfig:
     db_command_timeout: int
     db_acquire_timeout: int
 
-    # Background task processing
-    task_backend: str
-    task_backend_memory_batch_size: int
-    task_backend_memory_batch_interval: float
+    # Worker configuration (distributed task processing)
+    worker_enabled: bool
+    worker_id: str | None
+    worker_poll_interval_ms: int
+    worker_max_retries: int
+    worker_batch_size: int
+    worker_http_port: int
 
     # Reflect agent settings
     reflect_max_iterations: int
@@ -417,6 +441,15 @@ class HindsightConfig:
                 ENV_RETAIN_OBSERVATIONS_ASYNC, str(DEFAULT_RETAIN_OBSERVATIONS_ASYNC)
             ).lower()
             == "true",
+            # Mental models settings
+            enable_mental_models=os.getenv(ENV_ENABLE_MENTAL_MODELS, str(DEFAULT_ENABLE_MENTAL_MODELS)).lower()
+            == "true",
+            consolidation_similarity_threshold=float(
+                os.getenv(ENV_CONSOLIDATION_SIMILARITY_THRESHOLD, str(DEFAULT_CONSOLIDATION_SIMILARITY_THRESHOLD))
+            ),
+            consolidation_batch_size=int(
+                os.getenv(ENV_CONSOLIDATION_BATCH_SIZE, str(DEFAULT_CONSOLIDATION_BATCH_SIZE))
+            ),
             # Database migrations
             run_migrations_on_startup=os.getenv(ENV_RUN_MIGRATIONS_ON_STARTUP, "true").lower() == "true",
             # Database connection pool
@@ -424,14 +457,13 @@ class HindsightConfig:
             db_pool_max_size=int(os.getenv(ENV_DB_POOL_MAX_SIZE, str(DEFAULT_DB_POOL_MAX_SIZE))),
             db_command_timeout=int(os.getenv(ENV_DB_COMMAND_TIMEOUT, str(DEFAULT_DB_COMMAND_TIMEOUT))),
             db_acquire_timeout=int(os.getenv(ENV_DB_ACQUIRE_TIMEOUT, str(DEFAULT_DB_ACQUIRE_TIMEOUT))),
-            # Background task processing
-            task_backend=os.getenv(ENV_TASK_BACKEND, DEFAULT_TASK_BACKEND),
-            task_backend_memory_batch_size=int(
-                os.getenv(ENV_TASK_BACKEND_MEMORY_BATCH_SIZE, str(DEFAULT_TASK_BACKEND_MEMORY_BATCH_SIZE))
-            ),
-            task_backend_memory_batch_interval=float(
-                os.getenv(ENV_TASK_BACKEND_MEMORY_BATCH_INTERVAL, str(DEFAULT_TASK_BACKEND_MEMORY_BATCH_INTERVAL))
-            ),
+            # Worker configuration
+            worker_enabled=os.getenv(ENV_WORKER_ENABLED, str(DEFAULT_WORKER_ENABLED)).lower() == "true",
+            worker_id=os.getenv(ENV_WORKER_ID) or DEFAULT_WORKER_ID,
+            worker_poll_interval_ms=int(os.getenv(ENV_WORKER_POLL_INTERVAL_MS, str(DEFAULT_WORKER_POLL_INTERVAL_MS))),
+            worker_max_retries=int(os.getenv(ENV_WORKER_MAX_RETRIES, str(DEFAULT_WORKER_MAX_RETRIES))),
+            worker_batch_size=int(os.getenv(ENV_WORKER_BATCH_SIZE, str(DEFAULT_WORKER_BATCH_SIZE))),
+            worker_http_port=int(os.getenv(ENV_WORKER_HTTP_PORT, str(DEFAULT_WORKER_HTTP_PORT))),
             # Reflect agent settings
             reflect_max_iterations=int(os.getenv(ENV_REFLECT_MAX_ITERATIONS, str(DEFAULT_REFLECT_MAX_ITERATIONS))),
         )

hindsight-api/hindsight_api/engine/consolidation/__init__.py

@@ -0,0 +1,5 @@
+"""Consolidation engine for automatic learning creation from memories."""
+
+from .consolidator import run_consolidation_job
+
+__all__ = ["run_consolidation_job"]

hindsight-api/hindsight_api/engine/consolidation/consolidator.py

@@ -0,0 +1,842 @@
+"""Consolidation engine for automatic mental model creation from memories.
+
+The consolidation engine runs as a background job after retain operations complete.
+It processes new memories and either:
+- Creates new mental models from novel facts
+- Updates existing mental models when new evidence supports/contradicts/refines them
+
+Mental models are stored in memory_units with fact_type='mental_model' and include:
+- proof_count: Number of supporting memories
+- source_memory_ids: Array of memory UUIDs that contribute to this mental model
+- history: JSONB tracking changes over time
+"""
+
+import json
+import logging
+import time
+import uuid
+from datetime import datetime, timezone
+from typing import TYPE_CHECKING, Any
+
+from ..memory_engine import fq_table
+from ..retain import embedding_utils
+from .prompts import (
+    CONSOLIDATION_SYSTEM_PROMPT,
+    CONSOLIDATION_USER_PROMPT,
+)
+
+if TYPE_CHECKING:
+    from asyncpg import Connection
+
+    from ...api.http import RequestContext
+    from ..memory_engine import MemoryEngine
+
+logger = logging.getLogger(__name__)
+
+
+class ConsolidationPerfLog:
+    """Performance logging for consolidation operations."""
+
+    def __init__(self, bank_id: str):
+        self.bank_id = bank_id
+        self.start_time = time.time()
+        self.lines: list[str] = []
+        self.timings: dict[str, float] = {}
+
+    def log(self, message: str) -> None:
+        """Add a log line."""
+        self.lines.append(message)
+
+    def record_timing(self, key: str, duration: float) -> None:
+        """Record a timing measurement."""
+        if key in self.timings:
+            self.timings[key] += duration
+        else:
+            self.timings[key] = duration
+
+    def flush(self) -> None:
+        """Flush all log lines to the logger."""
+        total_time = time.time() - self.start_time
+        header = f"\n{'=' * 60}\nCONSOLIDATION for bank {self.bank_id}"
+        footer = f"{'=' * 60}\nCONSOLIDATION COMPLETE: {total_time:.3f}s total\n{'=' * 60}"
+
+        log_output = header + "\n" + "\n".join(self.lines) + "\n" + footer
+        logger.info(log_output)
+
+
+async def run_consolidation_job(
+    memory_engine: "MemoryEngine",
+    bank_id: str,
+    request_context: "RequestContext",
+) -> dict[str, Any]:
+    """
+    Run consolidation job for a bank.
+
+    This is called after retain operations to consolidate new memories into mental models.
+
+    Args:
+        memory_engine: MemoryEngine instance
+        bank_id: Bank identifier
+        request_context: Request context for authentication
+
+    Returns:
+        Dict with consolidation results
+    """
+    from ...config import get_config
+
+    config = get_config()
+    perf = ConsolidationPerfLog(bank_id)
+    max_memories_per_batch = config.consolidation_batch_size
+
+    # Check if consolidation is enabled
+    if not config.enable_mental_models:
+        logger.debug(f"Consolidation disabled for bank {bank_id}")
+        return {"status": "disabled", "bank_id": bank_id}
+
+    async with memory_engine._pool.acquire() as conn:
+        # Get bank profile and last_consolidated_at
+        t0 = time.time()
+        bank_row = await conn.fetchrow(
+            f"""
+            SELECT bank_id, name, mission, last_consolidated_at
+            FROM {fq_table("banks")}
+            WHERE bank_id = $1
+            """,
+            bank_id,
+        )
+
+        if not bank_row:
+            logger.warning(f"Bank {bank_id} not found for consolidation")
+            return {"status": "bank_not_found", "bank_id": bank_id}
+
+        mission = bank_row["mission"] or "General memory consolidation"
+        last_consolidated_at = bank_row["last_consolidated_at"]
+        perf.record_timing("fetch_bank", time.time() - t0)
+
+        # Fetch memories created after last_consolidated_at (exclude mental_model type)
+        t0 = time.time()
+        if last_consolidated_at:
+            memories = await conn.fetch(
+                f"""
+                SELECT id, text, fact_type, occurred_start, event_date, tags
+                FROM {fq_table("memory_units")}
+                WHERE bank_id = $1 AND created_at > $2
+                  AND fact_type IN ('experience', 'world')
+                ORDER BY created_at ASC
+                LIMIT $3
+                """,
+                bank_id,
+                last_consolidated_at,
+                max_memories_per_batch,
+            )
+        else:
+            memories = await conn.fetch(
+                f"""
+                SELECT id, text, fact_type, occurred_start, event_date, tags
+                FROM {fq_table("memory_units")}
+                WHERE bank_id = $1
+                  AND fact_type IN ('experience', 'world')
+                ORDER BY created_at ASC
+                LIMIT $2
+                """,
+                bank_id,
+                max_memories_per_batch,
+            )
+        perf.record_timing("fetch_memories", time.time() - t0)
+
+        if not memories:
+            logger.debug(f"No new memories to consolidate for bank {bank_id}")
+            # Update timestamp anyway to prevent reprocessing
+            await _update_last_consolidated_at(conn, bank_id)
+            return {"status": "no_new_memories", "bank_id": bank_id, "memories_processed": 0}
+
+        logger.info(
+            f"[CONSOLIDATION] bank={bank_id} memories={len(memories)} "
+            f"batch_size={max_memories_per_batch} since={last_consolidated_at or 'beginning'}"
+        )
+        perf.log(f"[1] Found {len(memories)} pending memories to consolidate")
+
+        # Process each memory sequentially
+        # Important: We process ALL pending memories before updating the watermark
+        # to avoid losing memories when many have the same timestamp
+        stats = {
+            "memories_processed": 0,
+            "mental_models_created": 0,
+            "mental_models_updated": 0,
+            "mental_models_merged": 0,
+            "actions_executed": 0,  # Total actions (can be > memories_processed due to multiple actions per fact)
+            "skipped": 0,
+        }
+
+        # Track processed memory IDs to avoid reprocessing
+        processed_ids: set[uuid.UUID] = set()
+        batch_num = 0
+
+        while memories:
+            batch_num += 1
+            batch_start = time.time()
+
+            for memory in memories:
+                if memory["id"] in processed_ids:
+                    continue
+
+                mem_start = time.time()
+                result = await _process_memory(
+                    conn=conn,
+                    memory_engine=memory_engine,
+                    bank_id=bank_id,
+                    memory=dict(memory),
+                    mission=mission,
+                    request_context=request_context,
+                    perf=perf,
+                )
+                mem_time = time.time() - mem_start
+                perf.record_timing("process_memory_total", mem_time)
+
+                processed_ids.add(memory["id"])
+                stats["memories_processed"] += 1
+
+                action = result.get("action")
+                if action == "created":
+                    stats["mental_models_created"] += 1
+                    stats["actions_executed"] += 1
+                elif action == "updated":
+                    stats["mental_models_updated"] += 1
+                    stats["actions_executed"] += 1
+                elif action == "merged":
+                    stats["mental_models_merged"] += 1
+                    stats["actions_executed"] += 1
+                elif action == "multiple":
+                    # Multiple actions from one fact (tag routing)
+                    stats["mental_models_created"] += result.get("created", 0)
+                    stats["mental_models_updated"] += result.get("updated", 0)
+                    stats["mental_models_merged"] += result.get("merged", 0)
+                    stats["actions_executed"] += result.get("total_actions", 0)
+                elif action == "skipped":
+                    stats["skipped"] += 1
+
+            batch_time = time.time() - batch_start
+            perf.log(
+                f"[2] Batch {batch_num}: {len(memories)} memories in {batch_time:.3f}s "
+                f"(avg {batch_time / len(memories):.3f}s/memory)"
+            )
+
+            # Fetch next batch of memories (excluding already processed)
+            t0 = time.time()
+            if last_consolidated_at:
+                memories = await conn.fetch(
+                    f"""
+                    SELECT id, text, fact_type, occurred_start, event_date, tags
+                    FROM {fq_table("memory_units")}
+                    WHERE bank_id = $1 AND created_at > $2
+                      AND fact_type IN ('experience', 'world')
+                      AND id != ALL($4)
+                    ORDER BY created_at ASC
+                    LIMIT $3
+                    """,
+                    bank_id,
+                    last_consolidated_at,
+                    max_memories_per_batch,
+                    list(processed_ids),
+                )
+            else:
+                memories = await conn.fetch(
+                    f"""
+                    SELECT id, text, fact_type, occurred_start, event_date, tags
+                    FROM {fq_table("memory_units")}
+                    WHERE bank_id = $1
+                      AND fact_type IN ('experience', 'world')
+                      AND id != ALL($3)
+                    ORDER BY created_at ASC
+                    LIMIT $2
+                    """,
+                    bank_id,
+                    max_memories_per_batch,
+                    list(processed_ids),
+                )
+            perf.record_timing("fetch_memories", time.time() - t0)
+
+        # Update last_consolidated_at only after ALL memories are processed
+        t0 = time.time()
+        await _update_last_consolidated_at(conn, bank_id)
+        perf.record_timing("update_watermark", time.time() - t0)
+
+        # Build summary
+        perf.log(
+            f"[3] Results: {stats['memories_processed']} memories → "
+            f"{stats['actions_executed']} actions "
+            f"({stats['mental_models_created']} created, "
+            f"{stats['mental_models_updated']} updated, "
+            f"{stats['mental_models_merged']} merged, "
+            f"{stats['skipped']} skipped)"
+        )
+
+        # Add timing breakdown
+        timing_parts = []
+        if "recall" in perf.timings:
+            timing_parts.append(f"recall={perf.timings['recall']:.3f}s")
+        if "llm" in perf.timings:
+            timing_parts.append(f"llm={perf.timings['llm']:.3f}s")
+        if "embedding" in perf.timings:
+            timing_parts.append(f"embedding={perf.timings['embedding']:.3f}s")
+        if "db_write" in perf.timings:
+            timing_parts.append(f"db_write={perf.timings['db_write']:.3f}s")
+
+        if timing_parts:
+            perf.log(f"[4] Timing breakdown: {', '.join(timing_parts)}")
+
+        perf.flush()
+
+        return {"status": "completed", "bank_id": bank_id, **stats}
+
+
+async def _update_last_consolidated_at(conn: "Connection", bank_id: str) -> None:
+    """Update the bank's last_consolidated_at timestamp."""
+    await conn.execute(
+        f"""
+        UPDATE {fq_table("banks")}
+        SET last_consolidated_at = $1
+        WHERE bank_id = $2
+        """,
+        datetime.now(timezone.utc),
+        bank_id,
+    )
+
+
+async def _process_memory(
+    conn: "Connection",
+    memory_engine: "MemoryEngine",
+    bank_id: str,
+    memory: dict[str, Any],
+    mission: str,
+    request_context: "RequestContext",
+    perf: ConsolidationPerfLog | None = None,
+) -> dict[str, Any]:
+    """
+    Process a single memory for consolidation using a SINGLE LLM call.
+
+    This function:
+    1. Finds related mental models (can be empty)
+    2. Uses ONE LLM call to extract durable knowledge AND decide on actions
+    3. Executes array of actions (can be multiple creates/updates)
+
+    The LLM handles all cases:
+    - No related models: returns create action(s) with extracted durable knowledge
+    - Related models exist: returns update/create actions based on tag routing
+    - Purely ephemeral fact: returns empty array (skip)
+
+    Returns:
+        Dict with action summary: created/updated/merged counts
+    """
+    fact_text = memory["text"]
+    memory_id = memory["id"]
+    fact_tags = memory.get("tags") or []
+
+    # Find related mental models using the full recall system (NO tag filtering)
+    t0 = time.time()
+    related_mental_models = await _find_related_mental_models(
+        conn=conn,
+        memory_engine=memory_engine,
+        bank_id=bank_id,
+        query=fact_text,
+        request_context=request_context,
+    )
+    if perf:
+        perf.record_timing("recall", time.time() - t0)
+
+    # Single LLM call handles ALL cases (with or without existing models)
+    t0 = time.time()
+    actions = await _consolidate_with_llm(
+        memory_engine=memory_engine,
+        fact_text=fact_text,
+        fact_tags=fact_tags,
+        mental_models=related_mental_models,  # Can be empty list
+        mission=mission,
+    )
+    if perf:
+        perf.record_timing("llm", time.time() - t0)
+
+    if not actions:
+        # LLM returned empty array - fact is purely ephemeral, skip
+        return {"action": "skipped", "reason": "no_durable_knowledge"}
+
+    # Execute all actions and collect results
+    results = []
+    for action in actions:
+        action_type = action.get("action")
+        if action_type == "update":
+            result = await _execute_update_action(
+                conn=conn,
+                memory_engine=memory_engine,
+                bank_id=bank_id,
+                memory_id=memory_id,
+                action=action,
+                mental_models=related_mental_models,
+                perf=perf,
+            )
+            results.append(result)
+        elif action_type == "create":
+            result = await _execute_create_action(
+                conn=conn,
+                memory_engine=memory_engine,
+                bank_id=bank_id,
+                memory_id=memory_id,
+                action=action,
+                event_date=memory.get("event_date"),
+                occurred_start=memory.get("occurred_start"),
+                perf=perf,
+            )
+            results.append(result)
+
+    if not results:
+        # No valid actions executed
+        return {"action": "skipped", "reason": "no_valid_actions"}
+
+    # Summarize results
+    created = sum(1 for r in results if r.get("action") == "created")
+    updated = sum(1 for r in results if r.get("action") == "updated")
+    merged = sum(1 for r in results if r.get("action") == "merged")
+
+    if len(results) == 1:
+        return results[0]
+
+    return {
+        "action": "multiple",
+        "created": created,
+        "updated": updated,
+        "merged": merged,
+        "total_actions": len(results),
+    }
+
+
+async def _execute_update_action(
+    conn: "Connection",
+    memory_engine: "MemoryEngine",
+    bank_id: str,
+    memory_id: uuid.UUID,
+    action: dict[str, Any],
+    mental_models: list[dict[str, Any]],
+    perf: ConsolidationPerfLog | None = None,
+) -> dict[str, Any]:
+    """
+    Execute an update action on an existing mental model.
+
+    Updates the mental model text, adds to history, and increments proof_count.
+    """
+    learning_id = action.get("learning_id")
+    new_text = action.get("text")
+    reason = action.get("reason", "Updated with new fact")
+
+    if not learning_id or not new_text:
+        return {"action": "skipped", "reason": "missing_learning_id_or_text"}
+
+    # Find the mental model
+    model = next((m for m in mental_models if str(m["id"]) == learning_id), None)
+    if not model:
+        return {"action": "skipped", "reason": "learning_not_found"}
+
+    # Build history entry
+    history = list(model.get("history", []))
+    history.append(
+        {
+            "previous_text": model["text"],
+            "changed_at": datetime.now(timezone.utc).isoformat(),
+            "reason": reason,
+            "source_memory_id": str(memory_id),
+        }
+    )
+
+    # Update source_memory_ids
+    source_ids = list(model.get("source_memory_ids", []))
+    source_ids.append(memory_id)
+
+    # Generate new embedding for updated text
+    t0 = time.time()
+    embeddings = await embedding_utils.generate_embeddings_batch(memory_engine.embeddings, [new_text])
+    embedding_str = str(embeddings[0]) if embeddings else None
+    if perf:
+        perf.record_timing("embedding", time.time() - t0)
+
+    # Update the mental model
+    t0 = time.time()
+    await conn.execute(
+        f"""
+        UPDATE {fq_table("memory_units")}
+        SET text = $1,
+            embedding = $2::vector,
+            history = $3,
+            source_memory_ids = $4,
+            proof_count = $5,
+            updated_at = now()
+        WHERE id = $6
+        """,
+        new_text,
+        embedding_str,
+        json.dumps(history),
+        source_ids,
+        len(source_ids),
+        uuid.UUID(learning_id),
+    )
+
+    # Create links from memory to mental model
+    await _create_memory_links(conn, memory_id, uuid.UUID(learning_id))
+    if perf:
+        perf.record_timing("db_write", time.time() - t0)
+
+    logger.debug(f"Updated mental model {learning_id} with memory {memory_id}")
+
+    return {"action": "updated", "mental_model_id": learning_id}
+
+
+async def _execute_create_action(
+    conn: "Connection",
+    memory_engine: "MemoryEngine",
+    bank_id: str,
+    memory_id: uuid.UUID,
+    action: dict[str, Any],
+    event_date: datetime | None = None,
+    occurred_start: datetime | None = None,
+    perf: ConsolidationPerfLog | None = None,
+) -> dict[str, Any]:
+    """
+    Execute a create action for a new mental model.
+
+    Creates a new mental model with the specified text and tags.
+    The text comes directly from the classify LLM - no second LLM call needed.
+    """
+    text = action.get("text")
+    tags = action.get("tags", [])
+
+    if not text:
+        return {"action": "skipped", "reason": "missing_text"}
+
+    # Use text directly from classify - skip the redundant LLM call
+    result = await _create_mental_model_directly(
+        conn=conn,
+        memory_engine=memory_engine,
+        bank_id=bank_id,
+        source_memory_id=memory_id,
+        mental_model_text=text,  # Text already processed by classify LLM
+        tags=tags,
+        event_date=event_date,
+        occurred_start=occurred_start,
+        perf=perf,
+    )
+
+    logger.debug(f"Created mental model {result.get('mental_model_id')} from memory {memory_id} (tags: {tags})")
+
+    return result
+
+
+async def _create_memory_links(
+    conn: "Connection",
+    memory_id: uuid.UUID,
+    mental_model_id: uuid.UUID,
+) -> None:
+    """
+    Create links between a source memory and its mental model.
+
+    This:
+    1. Creates bidirectional semantic links between memory and mental model
+    2. Copies existing memory_links from the source memory to the mental model
+    3. Copies entity links from the source memory to the mental model
+
+    This enables graph traversal to find related memories via their mental models.
+
+    Note: Uses EXISTS checks to handle the case where source memory was deleted
+    by a concurrent operation between fetching and link creation.
+    """
+    mu_table = fq_table("memory_units")
+    ml_table = fq_table("memory_links")
+    ue_table = fq_table("unit_entities")
+
+    # 1. Bidirectional link between memory and mental model
+    # Only insert if both units exist (handles concurrent deletion)
+    await conn.execute(
+        f"""
+        INSERT INTO {ml_table} (from_unit_id, to_unit_id, link_type, weight)
+        SELECT $1, $2, 'semantic', 1.0
+        WHERE EXISTS (SELECT 1 FROM {mu_table} WHERE id = $1)
+          AND EXISTS (SELECT 1 FROM {mu_table} WHERE id = $2)
+        ON CONFLICT DO NOTHING
+        """,
+        memory_id,
+        mental_model_id,
+    )
+    await conn.execute(
+        f"""
+        INSERT INTO {ml_table} (from_unit_id, to_unit_id, link_type, weight)
+        SELECT $1, $2, 'semantic', 1.0
+        WHERE EXISTS (SELECT 1 FROM {mu_table} WHERE id = $1)
+          AND EXISTS (SELECT 1 FROM {mu_table} WHERE id = $2)
+        ON CONFLICT DO NOTHING
+        """,
+        mental_model_id,
+        memory_id,
+    )
+
+    # 2. Copy outgoing memory_links from source memory to mental model
+    # If source memory links to X, mental model should also link to X
+    await conn.execute(
+        f"""
+        INSERT INTO {ml_table} (from_unit_id, to_unit_id, link_type, entity_id, weight)
+        SELECT $1, ml.to_unit_id, ml.link_type, ml.entity_id, ml.weight
+        FROM {ml_table} ml
+        WHERE ml.from_unit_id = $2 AND ml.to_unit_id != $1
+          AND EXISTS (SELECT 1 FROM {mu_table} WHERE id = $1)
+          AND EXISTS (SELECT 1 FROM {mu_table} WHERE id = ml.to_unit_id)
+        ON CONFLICT DO NOTHING
+        """,
+        mental_model_id,
+        memory_id,
+    )
+
+    # 3. Copy incoming memory_links from source memory to mental model
+    # If X links to source memory, X should also link to mental model
+    await conn.execute(
+        f"""
+        INSERT INTO {ml_table} (from_unit_id, to_unit_id, link_type, entity_id, weight)
+        SELECT ml.from_unit_id, $1, ml.link_type, ml.entity_id, ml.weight
+        FROM {ml_table} ml
+        WHERE ml.to_unit_id = $2 AND ml.from_unit_id != $1
+          AND EXISTS (SELECT 1 FROM {mu_table} WHERE id = $1)
+          AND EXISTS (SELECT 1 FROM {mu_table} WHERE id = ml.from_unit_id)
+        ON CONFLICT DO NOTHING
+        """,
+        mental_model_id,
+        memory_id,
+    )
+
+    # 4. Copy entity links from source memory to mental model
+    await conn.execute(
+        f"""
+        INSERT INTO {ue_table} (unit_id, entity_id)
+        SELECT $1, ue.entity_id
+        FROM {ue_table} ue
+        WHERE ue.unit_id = $2
+          AND EXISTS (SELECT 1 FROM {mu_table} WHERE id = $1)
+        ON CONFLICT DO NOTHING
+        """,
+        mental_model_id,
+        memory_id,
+    )
+
+
+async def _find_related_mental_models(
+    conn: "Connection",
+    memory_engine: "MemoryEngine",
+    bank_id: str,
+    query: str,
+    request_context: "RequestContext",
+) -> list[dict[str, Any]]:
+    """
+    Find mental models related to the given query using the full recall system.
+
+    IMPORTANT: We do NOT filter by tags here. Consolidation needs to see ALL
+    potentially related mental models regardless of scope, so the LLM can
+    decide on tag routing (same scope update vs cross-scope create).
+
+    This leverages:
+    - Semantic search (embedding similarity)
+    - BM25 text search (keyword matching)
+    - Entity-based retrieval (shared entities)
+    - Graph traversal (connected via entity links)
+
+    Returns:
+        List of related mental models with their tags for LLM tag routing
+    """
+    # Use recall to find related mental models
+    # NO tags parameter - we want ALL mental models regardless of scope
+    # Use low max_tokens since we only need mental models, not memories
+    recall_result = await memory_engine.recall_async(
+        bank_id=bank_id,
+        query=query,
+        max_tokens=5000,  # Token budget for mental models
+        fact_type=["mental_model"],  # Only retrieve mental models
+        request_context=request_context,
+        _quiet=True,  # Suppress logging
+        # NO tags parameter - intentionally get ALL mental models
+    )
+
+    # If no mental models returned, return empty list
+    # When fact_type=["mental_model"], results come back in `results` field
+    if not recall_result.results:
+        return []
+
+    # Trust recall's relevance filtering - fetch full data for each mental model
+    results = []
+    for mm in recall_result.results:
+        # Fetch full mental model data from DB to get history, source_memory_ids, tags
+        row = await conn.fetchrow(
+            f"""
+            SELECT id, text, proof_count, history, tags, source_memory_ids, created_at, updated_at
+            FROM {fq_table("memory_units")}
+            WHERE id = $1 AND bank_id = $2 AND fact_type = 'mental_model'
+            """,
+            uuid.UUID(mm.id),
+            bank_id,
+        )
+
+        if row:
+            history = row["history"]
+            if isinstance(history, str):
+                history = json.loads(history)
+            elif history is None:
+                history = []
+
+            results.append(
+                {
+                    "id": row["id"],
+                    "text": row["text"],
+                    "proof_count": row["proof_count"] or 1,
+                    "history": history,
+                    "tags": row["tags"] or [],  # Include tags for LLM tag routing
+                    "source_memory_ids": row["source_memory_ids"] or [],
+                    "similarity": 1.0,  # Retrieved via recall so assumed relevant
+                }
+            )
+
+    return results
+
+
+async def _consolidate_with_llm(
+    memory_engine: "MemoryEngine",
+    fact_text: str,
+    fact_tags: list[str],
+    mental_models: list[dict[str, Any]],
+    mission: str,
+) -> list[dict[str, Any]]:
+    """
+    Single LLM call to extract durable knowledge and decide on consolidation actions.
+
+    This handles ALL cases:
+    - No related mental models: extracts durable knowledge, returns create action
+    - Related models exist: compares and returns update/create actions
+    - Purely ephemeral fact: returns empty array
+
+    Returns:
+        List of actions, each being:
+        - {"action": "update", "learning_id": "uuid", "text": "...", "reason": "..."}
+        - {"action": "create", "tags": [...], "text": "...", "reason": "..."}
+        - [] if fact is purely ephemeral (no durable knowledge)
+    """
+    # Format mental models WITH their tags (or "None" if empty)
+    if mental_models:
+        mental_models_text = "\n".join(
+            f'- ID: {mm["id"]}, Tags: {json.dumps(mm["tags"])}, Text: "{mm["text"]}" (proof_count: {mm["proof_count"]})'
+            for mm in mental_models
+        )
+    else:
+        mental_models_text = "None (this is a new topic - create if fact contains durable knowledge)"
+
+    # Only include mission section if mission is set and not the default
+    mission_section = ""
+    if mission and mission != "General memory consolidation":
+        mission_section = f"""
+MISSION CONTEXT: {mission}
+
+Focus on DURABLE knowledge that serves this mission, not ephemeral state.
+"""
+
+    user_prompt = CONSOLIDATION_USER_PROMPT.format(
+        mission_section=mission_section,
+        fact_text=fact_text,
+        fact_tags=json.dumps(fact_tags),
+        mental_models_text=mental_models_text,
+    )
+
+    messages = [
+        {"role": "system", "content": CONSOLIDATION_SYSTEM_PROMPT},
+        {"role": "user", "content": user_prompt},
+    ]
+
+    try:
+        result = await memory_engine._llm_config.call(
+            messages=messages,
+            skip_validation=True,  # Raw JSON response
+            scope="consolidation",
+        )
+        # Parse JSON response - should be an array
+        if isinstance(result, str):
+            result = json.loads(result)
+        # Ensure result is a list
+        if isinstance(result, list):
+            return result
+        # Handle legacy single-action format for backward compatibility
+        if isinstance(result, dict):
+            if result.get("related_ids") and result.get("consolidated_text"):
+                # Convert old format to new format
+                return [
+                    {
+                        "action": "update",
+                        "learning_id": result["related_ids"][0],
+                        "text": result["consolidated_text"],
+                        "reason": result.get("reason", ""),
+                    }
+                ]
+            return []
+        return []
+    except Exception as e:
+        logger.warning(f"Error in consolidation LLM call: {e}")
+        return []
+
+
+async def _create_mental_model_directly(
+    conn: "Connection",
+    memory_engine: "MemoryEngine",
+    bank_id: str,
+    source_memory_id: uuid.UUID,
+    mental_model_text: str,
+    tags: list[str] | None = None,
+    event_date: datetime | None = None,
+    occurred_start: datetime | None = None,
+    perf: ConsolidationPerfLog | None = None,
+) -> dict[str, Any]:
+    """
+    Create a mental model directly with pre-processed text (no LLM call).
+
+    Used when the classify LLM has already provided the learning text.
+    This avoids the redundant second LLM call.
+    """
+    # Generate embedding for the mental model (convert to string for pgvector)
+    t0 = time.time()
+    embeddings = await embedding_utils.generate_embeddings_batch(memory_engine.embeddings, [mental_model_text])
+    embedding_str = str(embeddings[0]) if embeddings else None
+    if perf:
+        perf.record_timing("embedding", time.time() - t0)
+
+    # Create the mental model as a memory_unit
+    now = datetime.now(timezone.utc)
+    mm_event_date = event_date or now
+    mm_occurred_start = occurred_start or now
+    mm_tags = tags or []
+
+    t0 = time.time()
+    mental_model_id = uuid.uuid4()
+    row = await conn.fetchrow(
+        f"""
+        INSERT INTO {fq_table("memory_units")} (
+            id, bank_id, text, fact_type, embedding, proof_count, source_memory_ids, history,
+            tags, event_date, occurred_start
+        )
+        VALUES ($1, $2, $3, 'mental_model', $4::vector, 1, $5, '[]'::jsonb, $6, $7, $8)
+        RETURNING id
+        """,
+        mental_model_id,
+        bank_id,
+        mental_model_text,
+        embedding_str,
+        [source_memory_id],
+        mm_tags,
+        mm_event_date,
+        mm_occurred_start,
+    )
+
+    # Create links between memory and mental model (includes entity links, memory_links)
+    await _create_memory_links(conn, source_memory_id, mental_model_id)
+    if perf:
+        perf.record_timing("db_write", time.time() - t0)
+
+    logger.debug(f"Created mental model {mental_model_id} from memory {source_memory_id} (tags: {mm_tags})")
+
+    return {"action": "created", "mental_model_id": str(row["id"]), "tags": mm_tags}

hindsight-api/hindsight_api/engine/consolidation/prompts.py

@@ -0,0 +1,91 @@
+"""Prompts for the consolidation engine."""
+
+CONSOLIDATION_SYSTEM_PROMPT = """You are a memory consolidation system. Your job is to convert facts into durable knowledge (mental models) and merge with existing knowledge when appropriate.
+
+You must output ONLY valid JSON with no markdown formatting, no code blocks, and no additional text.
+
+## EXTRACT DURABLE KNOWLEDGE, NOT EPHEMERAL STATE
+Facts often describe events or actions. Extract the DURABLE KNOWLEDGE implied by the fact, not the transient state.
+
+Examples of extracting durable knowledge:
+- "User moved to Room 203" -> "Room 203 exists" (location exists, not where user is now)
+- "User visited Acme Corp at Room 105" -> "Acme Corp is located in Room 105"
+- "User took the elevator to floor 3" -> "Floor 3 is accessible by elevator"
+- "User met Sarah at the lobby" -> "Sarah can be found at the lobby"
+
+DO NOT track current user position/state as knowledge - that changes constantly.
+DO track permanent facts learned from the user's actions.
+
+## PRESERVE SPECIFIC DETAILS
+Keep names, locations, numbers, and other specifics. Do NOT:
+- Abstract into general principles
+- Generate business insights
+- Make knowledge generic
+
+GOOD examples:
+- Fact: "John likes pizza" -> "John likes pizza"
+- Fact: "Alice works at Google" -> "Alice works at Google"
+
+BAD examples:
+- "John likes pizza" -> "Understanding dietary preferences helps..." (TOO ABSTRACT)
+- "User is at Room 203" -> "User is currently at Room 203" (EPHEMERAL STATE)
+
+## MERGE RULES (when comparing to existing mental models):
+1. REDUNDANT: Same information worded differently → update existing
+2. CONTRADICTION: Opposite information about same topic → update with history (e.g., "used to X, now Y")
+3. UPDATE: New state replacing old state → update with history
+
+## TAG ROUTING RULES:
+Tags define visibility scopes. The fact and each mental model have tags (can be empty = global).
+
+| Fact Tags | Model Tags | Action |
+|-----------|------------|--------|
+| [alice] | [alice] | UPDATE the model (same scope) |
+| [alice] | [] | UPDATE the model (global absorbs all scopes) |
+| [alice] | [bob] | CREATE new untagged model (cross-scope insight) |
+| [] | [alice] | UPDATE the model (untagged facts can update any scope) |
+| [] | [] | UPDATE the model (global to global) |
+
+When NO existing model matches the fact's topic: CREATE new model with fact's tags.
+
+## MULTIPLE ACTIONS:
+One fact can trigger MULTIPLE actions. For example:
+- Update a scoped model [alice] about pizza preferences
+- AND update a global model [] about pizza in general
+
+Output an ARRAY of actions (can be empty, one, or many).
+
+## CRITICAL RULES:
+- NEVER merge facts about DIFFERENT people
+- NEVER merge unrelated topics (food preferences vs work vs hobbies)
+- When merging contradictions, capture the CHANGE (before → after)
+- Keep mental models focused on ONE specific topic per person
+- Cross-scope insights (alice's fact about bob's topic) become UNTAGGED (global)
+- The "text" field MUST contain durable knowledge, not ephemeral state"""
+
+CONSOLIDATION_USER_PROMPT = """Analyze this new fact and consolidate into knowledge.
+{mission_section}
+NEW FACT: {fact_text}
+FACT TAGS: {fact_tags}
+
+EXISTING MENTAL MODELS:
+{mental_models_text}
+
+Instructions:
+1. First, extract the DURABLE KNOWLEDGE from the fact (not ephemeral state like "user is at X")
+2. Then compare with existing mental models:
+   - If a model covers the same topic: UPDATE it with the new knowledge
+   - If no model covers the topic: CREATE a new one
+   - If fact is about different scope: apply tag routing rules
+
+Output JSON array of actions (ALWAYS an array, even for single action):
+[
+  {{"action": "update", "learning_id": "uuid", "text": "updated durable knowledge", "reason": "..."}},
+  {{"action": "create", "tags": ["tag"], "text": "new durable knowledge", "reason": "..."}}
+]
+
+If NO consolidation is needed (fact is purely ephemeral with no durable knowledge):
+[]
+
+If no models exist and fact contains durable knowledge:
+[{{"action": "create", "tags": {fact_tags}, "text": "durable knowledge text", "reason": "new topic"}}]"""

hindsight-api/hindsight_api/engine/cross_encoder.py

@@ -130,13 +130,28 @@ class LocalSTCrossEncoder(CrossEncoderModel):
                 "Install it with: pip install sentence-transformers"
             )
 
-        # Note: We use CPU even when GPU/MPS is available because:
-        # 1. The reranker model (MiniLM) is tiny (~22M params)
-        # 2. Batch sizes are small (~100-200 pairs)
-        # 3. Data transfer overhead to GPU outweighs compute benefit
-        # 4. CPU inference is actually faster for this workload
         logger.info(f"Reranker: initializing local provider with model {self.model_name}")
-        self._model = CrossEncoder(self.model_name)
+
+        # Determine device based on hardware availability.
+        # We always set low_cpu_mem_usage=False to prevent lazy loading (meta tensors)
+        # which can cause issues when accelerate is installed but no GPU is available.
+        # Note: We do NOT use device_map because CrossEncoder internally calls .to(device)
+        # after loading, which conflicts with accelerate's device_map handling.
+        import torch
+
+        # Check for GPU (CUDA) or Apple Silicon (MPS)
+        has_gpu = torch.cuda.is_available() or (hasattr(torch.backends, "mps") and torch.backends.mps.is_available())
+
+        if has_gpu:
+            device = None  # Let sentence-transformers auto-detect GPU/MPS
+        else:
+            device = "cpu"
+
+        self._model = CrossEncoder(
+            self.model_name,
+            device=device,
+            model_kwargs={"low_cpu_mem_usage": False},
+        )
 
         # Initialize shared executor (limited workers naturally limits concurrency)
         if LocalSTCrossEncoder._executor is None:

hindsight-api/hindsight_api/engine/directives/__init__.py

@@ -0,0 +1,5 @@
+"""Directives module for hard rules injected into prompts."""
+
+from .models import Directive
+
+__all__ = ["Directive"]

hindsight-api/hindsight_api/engine/directives/models.py

@@ -0,0 +1,37 @@
+"""Pydantic models for directives."""
+
+from datetime import datetime, timezone
+from uuid import UUID
+
+from pydantic import BaseModel, Field
+
+
+class Directive(BaseModel):
+    """A directive is a hard rule injected into prompts.
+
+    Directives are user-defined rules that guide agent behavior. Unlike mental models
+    which are automatically consolidated from memories, directives are explicit
+    instructions that are always included in relevant prompts.
+
+    Examples:
+    - "Always respond in formal English"
+    - "Never share personal data with third parties"
+    - "Prefer conservative investment recommendations"
+    """
+
+    id: UUID = Field(description="Unique identifier")
+    bank_id: str = Field(description="Bank this directive belongs to")
+    name: str = Field(description="Human-readable name")
+    content: str = Field(description="The directive text to inject into prompts")
+    priority: int = Field(default=0, description="Higher priority directives are injected first")
+    is_active: bool = Field(default=True, description="Whether this directive is currently active")
+    tags: list[str] = Field(default_factory=list, description="Tags for filtering")
+    created_at: datetime = Field(
+        default_factory=lambda: datetime.now(timezone.utc), description="When this directive was created"
+    )
+    updated_at: datetime = Field(
+        default_factory=lambda: datetime.now(timezone.utc), description="When this directive was last updated"
+    )
+
+    class Config:
+        from_attributes = True

hindsight-api/hindsight_api/engine/embeddings.py

@@ -128,11 +128,24 @@ class LocalSTEmbeddings(Embeddings):
             )
 
         logger.info(f"Embeddings: initializing local provider with model {self.model_name}")
-        # Disable lazy loading (meta tensors) which causes issues with newer transformers/accelerate
-        # Setting low_cpu_mem_usage=False and device_map=None ensures tensors are fully materialized
+
+        # Determine device based on hardware availability.
+        # We always set low_cpu_mem_usage=False to prevent lazy loading (meta tensors)
+        # which can cause issues when accelerate is installed but no GPU is available.
+        import torch
+
+        # Check for GPU (CUDA) or Apple Silicon (MPS)
+        has_gpu = torch.cuda.is_available() or (hasattr(torch.backends, "mps") and torch.backends.mps.is_available())
+
+        if has_gpu:
+            device = None  # Let sentence-transformers auto-detect GPU/MPS
+        else:
+            device = "cpu"
+
         self._model = SentenceTransformer(
             self.model_name,
-            model_kwargs={"low_cpu_mem_usage": False, "device_map": None},
+            device=device,
+            model_kwargs={"low_cpu_mem_usage": False},
         )
 
         self._dimension = self._model.get_sentence_embedding_dimension()

hindsight-api/hindsight_api/engine/mental_models/__init__.py

@@ -1,16 +1,12 @@
 """
 Mental models module for Hindsight.
 
-Mental models are synthesized summaries that represent understanding. They come
-in different subtypes based on how they were created:
+Mental models contain directives - hard rules that are injected into reflect prompts.
+Directives are user-defined and their observations are user-provided (not LLM-generated).
 
-- Structural: Derived from the bank's mission (e.g., "Be a PM for engineering team")
-  These are created upfront based on what any agent with this role would need.
-
-- Emergent: Discovered from data patterns (named entities, temporal clusters, etc.)
-  These surface organically as facts are retained.
-
-- Pinned: User-defined models that persist across refreshes.
+Other types of consolidated knowledge are handled by:
+- Learnings: Automatic bottom-up consolidation from facts
+- Pinned Reflections: User-curated living documents
 """
 
 from .models import MentalModel, MentalModelSubtype

hindsight-api/hindsight_api/engine/mental_models/emergent.py

@@ -1,311 +0,0 @@
-"""
-Emergent mental model detection and promotion.
-
-Emergent models are discovered from data patterns:
-- Named entity extraction (people, projects, systems)
-- Temporal clustering (events with multiple references)
-- Causal patterns ("Because X, we do Y")
-- Behavioral anchors ("After X, we started Y")
-- Reference frequency (anything mentioned repeatedly)
-
-When a pattern is detected, it goes through a mission filter to check relevance,
-and if relevant, is promoted to a mental model.
-"""
-
-import logging
-from typing import TYPE_CHECKING
-
-from pydantic import BaseModel, Field
-
-from .models import EmergentCandidate
-
-if TYPE_CHECKING:
-    from ..llm_wrapper import LLMConfig
-
-logger = logging.getLogger(__name__)
-
-
-class MissionFilterCandidate(BaseModel):
-    """Result of mission filtering for a single candidate."""
-
-    name: str
-    promote: bool = Field(description="True if this is a specific named entity worth tracking")
-    reason: str = Field(description="Brief explanation for the decision")
-
-
-class MissionFilterResponse(BaseModel):
-    """Response from LLM for mission filtering."""
-
-    candidates: list[MissionFilterCandidate] = Field(description="Filtering decision for each candidate")
-
-
-def build_mission_filter_prompt(mission: str, candidates: list[EmergentCandidate]) -> str:
-    """Build the prompt for filtering candidates by mission relevance."""
-    candidate_list = "\n".join(
-        [f"- {c.name} (mentions: {c.mention_count}, method: {c.detection_method})" for c in candidates]
-    )
-
-    return f"""Filter these detected entities. For each one, decide: promote=true or promote=false.
-
-MISSION: {mission}
-
-DETECTED ENTITIES:
-{candidate_list}
-
-=== DECISION RULES ===
-
-Set promote=true ONLY for specific, named entities:
-- Person names: "John", "Maria", "Alice Chen", "Dr. Smith"
-- Named organizations: "Google", "Acme Corp", "Frontend Team"
-- Named places: "Central Park Zoo", "NYC Office", "Building A"
-- Named projects: "Project Phoenix", "Auth Service v2"
-
-Set promote=false for EVERYTHING ELSE, including:
-- Common English words: user, support, help, family, kids, parents, friends, people, team, photo, nature, park, office, home, work, school, joy, love, hope, fear, anger, gratitude, kindness, passion, motivation, inspiration, encouragement, positivity, energy, community, connection, commitment, collaboration, growth, impact, difference, success, progress, change, education, volunteering, veterans, homeless, shelter, meeting, project, system, process, event
-- Generic categories (even capitalized): Users, Customers, Team, Family, Kids, Veterans, Community
-- Abstract concepts: motivation, inspiration, gratitude, commitment, resilience
-
-THE TEST: Is this a specific name you'd find in a contact list or org chart?
-- "John" → YES (promote=true)
-- "kids" → NO (promote=false)
-- "community" → NO (promote=false)
-- "Maria" → YES (promote=true)
-- "park" → NO (promote=false)
-
-When in doubt, set promote=false."""
-
-
-def get_mission_filter_system_message() -> str:
-    """System message for mission filtering."""
-    return """You filter entities for promotion. Output JSON with 'candidates' array.
-
-Rules:
-- promote=true ONLY for specific names (people, organizations, named places/projects)
-- promote=false for common words, generic categories, abstract concepts
-
-Examples:
-- "John" → promote=true (person name)
-- "kids" → promote=false (generic category)
-- "community" → promote=false (abstract concept)
-- "Google" → promote=true (organization name)
-- "motivation" → promote=false (abstract concept)
-
-When in doubt, promote=false. Most entities should be rejected."""
-
-
-async def filter_candidates_by_mission(
-    llm_config: "LLMConfig",
-    mission: str,
-    candidates: list[EmergentCandidate],
-) -> list[EmergentCandidate]:
-    """
-    Filter emergent candidates to keep only specific, named entities.
-
-    Args:
-        llm_config: LLM configuration
-        mission: The bank's mission (used for context)
-        candidates: List of detected candidates
-
-    Returns:
-        Filtered list of candidates that are specific named entities
-    """
-    if not candidates:
-        return []
-
-    if not mission:
-        # No mission = no filtering, keep all candidates
-        logger.debug("[EMERGENT] No mission set, skipping filter")
-        return candidates
-
-    prompt = build_mission_filter_prompt(mission, candidates)
-
-    try:
-        result = await llm_config.call(
-            messages=[
-                {"role": "system", "content": get_mission_filter_system_message()},
-                {"role": "user", "content": prompt},
-            ],
-            response_format=MissionFilterResponse,
-            scope="mental_model_mission_filter",
-        )
-
-        # Build name -> promote map
-        promote_map = {c.name: c.promote for c in result.candidates}
-
-        # Filter candidates
-        filtered = []
-        for candidate in candidates:
-            if candidate.name in promote_map:
-                if promote_map[candidate.name]:
-                    filtered.append(candidate)
-                    logger.debug(f"[EMERGENT] Promoting '{candidate.name}'")
-                else:
-                    logger.debug(f"[EMERGENT] Rejecting '{candidate.name}'")
-            else:
-                # Candidate not in response - reject by default
-                logger.debug(f"[EMERGENT] '{candidate.name}' not in response, rejecting")
-
-        logger.info(f"[EMERGENT] Mission filter: {len(filtered)}/{len(candidates)} candidates promoted")
-        return filtered
-
-    except Exception as e:
-        logger.warning(f"[EMERGENT] Mission filter failed, rejecting all candidates: {e}")
-        return []
-
-
-async def evaluate_emergent_models(
-    llm_config: "LLMConfig",
-    models: list[dict],
-) -> list[str]:
-    """
-    Evaluate existing emergent models to check if they should be kept.
-
-    This re-evaluates emergent models using the same filtering criteria
-    as new candidates. Models that are generic/abstract will be removed.
-
-    Args:
-        llm_config: LLM configuration
-        models: List of existing emergent model dicts with 'name', 'id'
-
-    Returns:
-        List of model IDs that should be REMOVED (no longer valid)
-    """
-    if not models:
-        return []
-
-    # Convert existing models to candidates for evaluation
-    candidates = [
-        EmergentCandidate(
-            name=m["name"],
-            detection_method="existing_emergent_model",
-            mention_count=0,
-        )
-        for m in models
-    ]
-
-    # Build a simple prompt for re-evaluation
-    names_list = "\n".join([f"- {m['name']}" for m in models])
-    prompt = f"""Re-evaluate these existing mental models. For each one, decide: promote=true (keep) or promote=false (remove).
-
-EXISTING MODELS:
-{names_list}
-
-=== DECISION RULES ===
-
-Set promote=true ONLY for specific, named entities:
-- Person names: "John", "Maria", "Alice Chen", "Dr. Smith"
-- Named organizations: "Google", "Acme Corp", "Frontend Team"
-- Named places: "Central Park Zoo", "NYC Office", "Building A"
-- Named projects: "Project Phoenix", "Auth Service v2"
-
-Set promote=false for EVERYTHING ELSE, including:
-- Common English words: user, support, help, family, kids, parents, friends, people, team, photo, nature, park, office, home, work, school, joy, love, hope, fear, anger, gratitude, kindness, passion, motivation, inspiration, encouragement, positivity, energy, community, connection, commitment, collaboration, growth, impact, difference, success, progress, change, education, volunteering, veterans, homeless, shelter, meeting, project, system, process, event
-- Generic categories (even capitalized): Users, Customers, Team, Family, Kids, Veterans, Community
-- Abstract concepts: motivation, inspiration, gratitude, commitment, resilience
-
-THE TEST: Is this a specific name you'd find in a contact list or org chart?
-- "John" → YES (promote=true)
-- "kids" → NO (promote=false)
-- "community" → NO (promote=false)
-
-When in doubt, set promote=false."""
-
-    try:
-        result = await llm_config.call(
-            messages=[
-                {"role": "system", "content": get_mission_filter_system_message()},
-                {"role": "user", "content": prompt},
-            ],
-            response_format=MissionFilterResponse,
-            scope="mental_model_emergent_evaluation",
-        )
-
-        # Build name -> promote map
-        promote_map = {c.name: c.promote for c in result.candidates}
-
-        # Find models to remove
-        models_to_remove = []
-        for model in models:
-            name = model["name"]
-            if name in promote_map:
-                if not promote_map[name]:
-                    models_to_remove.append(model["id"])
-                else:
-                    logger.debug(f"[EMERGENT] Keeping '{name}'")
-            else:
-                # Model not in response - remove to be safe
-                logger.info(f"[EMERGENT] '{name}' not in evaluation response, marking for removal")
-                models_to_remove.append(model["id"])
-
-        logger.info(f"[EMERGENT] Evaluation: {len(models_to_remove)}/{len(models)} emergent models marked for removal")
-        return models_to_remove
-
-    except Exception as e:
-        logger.warning(f"[EMERGENT] Evaluation failed, keeping all models: {e}")
-        return []
-
-
-async def detect_entity_candidates(
-    pool,
-    bank_id: str,
-    min_mentions: int = 5,
-    top_percent: int = 20,
-) -> list[EmergentCandidate]:
-    """
-    Detect entities that are candidates for promotion to mental models.
-
-    Args:
-        pool: Database connection pool
-        bank_id: Bank identifier
-        min_mentions: Minimum mention count to consider
-        top_percent: Only consider top X% by mention count
-
-    Returns:
-        List of entity candidates
-    """
-    from ..db_utils import acquire_with_retry
-    from ..memory_engine import fq_table
-
-    candidates = []
-
-    async with acquire_with_retry(pool) as conn:
-        # Get entities that meet criteria and don't already have mental models
-        rows = await conn.fetch(
-            f"""
-            WITH ranked AS (
-                SELECT
-                    e.id,
-                    e.canonical_name,
-                    e.mention_count,
-                    PERCENT_RANK() OVER (ORDER BY e.mention_count DESC) as rank_pct
-                FROM {fq_table("entities")} e
-                LEFT JOIN {fq_table("mental_models")} mm
-                    ON mm.entity_id = e.id AND mm.bank_id = e.bank_id
-                WHERE e.bank_id = $1
-                  AND e.mention_count >= $2
-                  AND mm.id IS NULL  -- Not already a mental model
-            )
-            SELECT id, canonical_name, mention_count
-            FROM ranked
-            WHERE rank_pct <= $3
-            ORDER BY mention_count DESC
-            LIMIT 50
-            """,
-            bank_id,
-            min_mentions,
-            top_percent / 100.0,
-        )
-
-        for row in rows:
-            candidates.append(
-                EmergentCandidate(
-                    name=row["canonical_name"],
-                    detection_method="named_entity_extraction",
-                    mention_count=row["mention_count"],
-                    entity_id=str(row["id"]),
-                    relevance_score=0.0,
-                )
-            )
-
-    logger.debug(f"[EMERGENT] Detected {len(candidates)} entity candidates")
-    return candidates

hindsight-api/hindsight_api/engine/mental_models/models.py

@@ -9,12 +9,14 @@ from pydantic import BaseModel, Field
 
 
 class MentalModelSubtype(str, Enum):
-    """Subtype of mental model - how it was created."""
+    """Subtype of mental model.
+
+    Currently only DIRECTIVE is supported. Other types of consolidated knowledge
+    are handled by:
+    - Learnings: Automatic bottom-up consolidation from facts
+    - Pinned Reflections: User-curated living documents
+    """
 
-    STRUCTURAL = "structural"  # Derived from mission, created upfront
-    EMERGENT = "emergent"  # Discovered from data patterns
-    LEARNED = "learned"  # Formed through reflection
-    PINNED = "pinned"  # User-defined topic, observations LLM-generated
     DIRECTIVE = "directive"  # User-defined hard rules, observations user-provided
 
 
@@ -49,50 +51,3 @@ class MentalModel(BaseModel):
     created_at: datetime = Field(
         default_factory=lambda: datetime.now(timezone.utc), description="When this model was created"
     )
-
-
-class StructuralModelTemplate(BaseModel):
-    """
-    A template for a structural mental model.
-
-    Generated by LLM based on the bank's mission. Represents what any agent
-    with this role would need to track.
-    """
-
-    id: str = Field(default="", description="Existing model ID to keep, or empty for new models")
-    name: str = Field(description="Human-readable name")
-    description: str = Field(description="What this model should track")
-    initial_probes: list[str] = Field(default_factory=list, description="Initial search queries to populate this model")
-
-
-class StructuralModelDerivationResponse(BaseModel):
-    """Response from LLM for structural model derivation."""
-
-    templates: list[StructuralModelTemplate] = Field(description="Structural model templates derived from the mission")
-
-
-class EmergentCandidate(BaseModel):
-    """
-    A candidate for promotion to emergent mental model.
-
-    Detected through pattern analysis of facts.
-    """
-
-    name: str = Field(description="Name of the detected pattern/entity")
-    detection_method: str = Field(description="How this candidate was detected")
-    mention_count: int = Field(default=0, description="How many times referenced")
-    entity_id: str | None = Field(default=None, description="Entity ID if detected as entity")
-    relevance_score: float = Field(default=0.0, description="Score from mission filter (0-1)")
-
-
-class ResearchResult(BaseModel):
-    """
-    Result from the research endpoint.
-
-    Contains the answer along with the mental models and facts used.
-    """
-
-    answer: str = Field(description="The synthesized answer")
-    mental_models_used: list[str] = Field(default_factory=list, description="IDs of mental models that contributed")
-    facts_used: list[str] = Field(default_factory=list, description="Fact IDs that contributed")
-    question_type: str | None = Field(default=None, description="Detected question type (WHO, WHAT, HOW, etc.)")

hindsight-api/hindsight_api/engine/mental_models/structural.py

@@ -1,228 +0,0 @@
-"""
-Structural mental model derivation from bank mission.
-
-Structural models are derived from the bank's mission - they represent what
-any agent with this role would need to track. For example:
-
-Mission: "Be a PM for engineering team"
-Structural models:
-  - Team Structure (who's on the team, roles)
-  - Project Overview (current projects, status)
-  - Processes (how releases work, how decisions are made)
-  - Key Systems (what we own, dependencies)
-"""
-
-import logging
-from typing import TYPE_CHECKING
-
-from pydantic import BaseModel, Field
-
-from .models import StructuralModelTemplate
-
-if TYPE_CHECKING:
-    from ..llm_wrapper import LLMConfig
-
-logger = logging.getLogger(__name__)
-
-
-class StructuralDerivationResponse(BaseModel):
-    """Response from LLM for structural model derivation."""
-
-    templates: list[StructuralModelTemplate] = Field(description="Structural model templates derived from the mission")
-
-
-class StructuralRelevanceResult(BaseModel):
-    """Result of evaluating a structural model's relevance to the mission."""
-
-    name: str
-    relevant: bool
-    reason: str
-
-
-class StructuralRelevanceResponse(BaseModel):
-    """Response from LLM for structural model relevance evaluation."""
-
-    models: list[StructuralRelevanceResult] = Field(description="Relevance evaluation for each model")
-
-
-def build_structural_derivation_prompt(mission: str, existing_models: list[dict] | None = None) -> str:
-    """Build the prompt for deriving structural models from a mission."""
-    existing_section = ""
-    if existing_models:
-        model_list = "\n".join([f"- id='{m['id']}' name='{m['name']}': {m['description']}" for m in existing_models])
-        existing_section = f"""
-EXISTING STRUCTURAL MODELS:
-{model_list}
-
-IMPORTANT: If keeping an existing model, you MUST return its EXACT 'id' value.
-Models not included in your output will be REMOVED.
-"""
-
-    return f"""Given this agent mission, identify the KEY THINGS to track to achieve it.
-
-MISSION: {mission}
-{existing_section}
-IMPORTANT CONSTRAINTS:
-- Return 0-3 structural models MAXIMUM (less is better!)
-- Only include models for SPECIFIC, CONCRETE things the agent needs to track
-- Each model must be DIRECTLY tied to achieving the mission
-- If the mission is simple, return 0 models (empty array is fine)
-- If existing models are provided and you want to keep one, use its EXACT id
-- Do NOT create near-duplicates (e.g., don't create "topic-map" if "topic-connections" exists)
-
-GOOD examples (specific, actionable):
-- Mission: "Be a PM for engineering team" → "Team Members" (track who's on the team)
-- Mission: "Track customer feedback" → "Customer Issues" (track specific complaints/requests)
-- Mission: "Manage project X" → "Project X Milestones" (track progress)
-
-BAD examples (too generic, don't create these):
-- "Processes", "Workflows", "Key Systems", "Important Events"
-- "Communication", "Collaboration", "Progress", "Status"
-- Generic role-based models not tied to the specific mission
-
-For each model:
-1. id: Use EXACT existing id if keeping a model, or leave empty for new models
-2. name: Short, specific name (e.g., "Team Members", "Sprint Goals")
-3. description: One line describing what to track
-4. initial_probes: 2-3 search queries to find relevant information
-
-Return ONLY the models that should exist. Existing models not in your output will be deleted."""
-
-
-def get_structural_derivation_system_message() -> str:
-    """System message for structural model derivation."""
-    return """You identify the key things to track for a mission. Be VERY selective.
-
-Rules:
-- Maximum 3 models (prefer fewer)
-- Only SPECIFIC, CONCRETE things - not generic categories
-- Each must DIRECTLY help achieve the mission
-- Empty array is valid if no models are truly needed
-- If existing models are shown and you want to keep one, return its EXACT id
-- Never create duplicates - if a similar model exists, keep the existing one
-
-Output JSON with 'templates' array (can be empty)."""
-
-
-def _normalize_id(text: str) -> str:
-    """Normalize a string to a canonical form for comparison.
-
-    Removes common suffixes, pluralization, and normalizes separators.
-    """
-    # Lowercase and normalize separators
-    normalized = text.lower().replace(" ", "-").replace("_", "-")
-
-    # Remove common suffixes that indicate the same concept
-    suffixes_to_remove = ["-map", "-list", "-overview", "-tracker", "-s"]
-    for suffix in suffixes_to_remove:
-        if normalized.endswith(suffix) and len(normalized) > len(suffix):
-            normalized = normalized[: -len(suffix)]
-
-    return normalized
-
-
-def _find_similar_existing_id(new_id: str, existing_models: list[dict]) -> str | None:
-    """Find an existing model ID that is similar to the new ID.
-
-    Returns the existing ID if a similar one is found, None otherwise.
-    """
-    if not existing_models:
-        return None
-
-    new_normalized = _normalize_id(new_id)
-
-    for model in existing_models:
-        existing_id = model.get("id", "")
-        existing_normalized = _normalize_id(existing_id)
-
-        # Check if one is a prefix of the other (normalized)
-        if new_normalized.startswith(existing_normalized) or existing_normalized.startswith(new_normalized):
-            return existing_id
-
-        # Check if they're the same when normalized
-        if new_normalized == existing_normalized:
-            return existing_id
-
-    return None
-
-
-async def derive_structural_models(
-    llm_config: "LLMConfig",
-    mission: str,
-    existing_models: list[dict] | None = None,
-) -> tuple[list[StructuralModelTemplate], list[str]]:
-    """
-    Derive structural model templates from a bank's mission.
-
-    This combines derivation and evaluation in one call. The LLM sees existing
-    models and decides which to keep. Any existing model not in the output
-    will be marked for removal.
-
-    Args:
-        llm_config: LLM configuration for calling the model
-        mission: The bank's mission (e.g., "Be a PM for engineering team")
-        existing_models: Optional list of existing model dicts with 'name', 'description', 'id'
-
-    Returns:
-        Tuple of (templates to create/keep, IDs of existing models to remove)
-
-    Raises:
-        Exception: If LLM call fails
-    """
-    prompt = build_structural_derivation_prompt(mission, existing_models)
-
-    result = await llm_config.call(
-        messages=[
-            {"role": "system", "content": get_structural_derivation_system_message()},
-            {"role": "user", "content": prompt},
-        ],
-        response_format=StructuralDerivationResponse,
-        scope="mental_model_structural_derivation",
-    )
-
-    templates = result.templates
-    logger.info(f"[STRUCTURAL] LLM returned {len(templates)} structural models")
-
-    # Build set of existing IDs for quick lookup
-    existing_ids = {m["id"] for m in existing_models} if existing_models else set()
-
-    # Process templates: validate IDs, deduplicate, assign stable IDs
-    processed_templates: list[StructuralModelTemplate] = []
-    kept_existing_ids: set[str] = set()
-
-    for template in templates:
-        # If LLM returned an ID, check if it's a valid existing ID
-        if template.id and template.id in existing_ids:
-            # LLM is keeping an existing model
-            kept_existing_ids.add(template.id)
-            processed_templates.append(template)
-            logger.info(f"[STRUCTURAL] Keeping existing model: {template.id}")
-        else:
-            # New model or LLM didn't return a valid ID
-            # Generate ID from name
-            generated_id = template.name.lower().replace(" ", "-").replace("_", "-")
-
-            # Check for similar existing models to prevent near-duplicates
-            similar_id = _find_similar_existing_id(generated_id, existing_models)
-            if similar_id and similar_id not in kept_existing_ids:
-                # Use the existing similar model instead of creating a new one
-                logger.info(f"[STRUCTURAL] Detected near-duplicate: '{generated_id}' matches existing '{similar_id}'")
-                template.id = similar_id
-                kept_existing_ids.add(similar_id)
-            else:
-                template.id = generated_id
-
-            processed_templates.append(template)
-
-    # Find existing models to remove (not kept in LLM output)
-    models_to_remove = []
-    if existing_models:
-        for model in existing_models:
-            if model["id"] not in kept_existing_ids:
-                logger.info(f"[STRUCTURAL] Marking '{model['name']}' (id={model['id']}) for removal")
-                models_to_remove.append(model["id"])
-
-    if models_to_remove:
-        logger.info(f"[STRUCTURAL] {len(models_to_remove)} existing models will be removed")
-
-    return processed_templates, models_to_remove

hindsight-api/hindsight_api/engine/reflect/agent.py

@@ -1,5 +1,10 @@
 """
 Reflect agent - agentic loop for reflection with native tool calling.
+
+Uses hierarchical retrieval:
+1. search_reflections - User-curated summaries (highest quality)
+2. search_mental_models - Consolidated knowledge with freshness
+3. recall - Raw facts as ground truth
 """
 
 import asyncio
@@ -8,7 +13,7 @@ import logging
 import time
 from typing import TYPE_CHECKING, Any, Awaitable, Callable
 
-from .models import DirectiveInfo, LLMCall, MentalModelInput, ReflectAgentResult, ToolCall
+from .models import DirectiveInfo, LLMCall, ReflectAgentResult, ToolCall
 from .prompts import FINAL_SYSTEM_PROMPT, _extract_directive_rules, build_final_prompt, build_system_prompt_for_tools
 from .tools_schema import get_reflect_tools
 
@@ -46,6 +51,32 @@ logger = logging.getLogger(__name__)
 DEFAULT_MAX_ITERATIONS = 10
 
 
+def _normalize_tool_name(name: str) -> str:
+    """Normalize tool name from various LLM output formats.
+
+    Some LLMs output tool names in non-standard formats:
+    - 'functions.done' (OpenAI-style prefix)
+    - 'call=functions.done' (some models)
+    - 'call=done' (some models)
+
+    Returns the normalized tool name (e.g., 'done', 'recall', etc.)
+    """
+    # Handle 'call=functions.name' or 'call=name' format
+    if name.startswith("call="):
+        name = name[len("call=") :]
+
+    # Handle 'functions.name' format
+    if name.startswith("functions."):
+        name = name[len("functions.") :]
+
+    return name
+
+
+def _is_done_tool(name: str) -> bool:
+    """Check if the tool name represents the 'done' tool."""
+    return _normalize_tool_name(name) == "done"
+
+
 async def _generate_structured_output(
     answer: str,
     response_schema: dict,
@@ -153,10 +184,10 @@ async def run_reflect_agent(
     bank_id: str,
     query: str,
     bank_profile: dict[str, Any],
-    lookup_fn: Callable[[str | None], Awaitable[dict[str, Any]]],
+    search_reflections_fn: Callable[[str, int], Awaitable[dict[str, Any]]],
+    search_mental_models_fn: Callable[[str, int], Awaitable[dict[str, Any]]],
     recall_fn: Callable[[str, int], Awaitable[dict[str, Any]]],
     expand_fn: Callable[[list[str], str], Awaitable[dict[str, Any]]],
-    learn_fn: Callable[[MentalModelInput], Awaitable[dict[str, Any]]] | None = None,
     context: str | None = None,
     max_iterations: int = DEFAULT_MAX_ITERATIONS,
     max_tokens: int | None = None,
@@ -166,19 +197,20 @@ async def run_reflect_agent(
     """
     Execute the reflect agent loop using native tool calling.
 
-    The agent iteratively calls tools to gather information and learn,
-    then provides a final answer via the done() tool.
+    The agent uses hierarchical retrieval:
+    1. search_reflections - User-curated summaries (try first)
+    2. search_mental_models - Consolidated knowledge with freshness
+    3. recall - Raw facts as ground truth
 
     Args:
         llm_config: LLM provider for agent calls
         bank_id: Bank identifier
         query: Question to answer
         bank_profile: Bank profile with name and mission
-        lookup_fn: Tool callback for lookup (model_id) -> result
+        search_reflections_fn: Tool callback for searching reflections (query, max_results) -> result
+        search_mental_models_fn: Tool callback for searching mental models (query, max_results) -> result
         recall_fn: Tool callback for recall (query, max_tokens) -> result
-        expand_fn: Tool callback for expand (memory_id, depth) -> result
-        learn_fn: Optional tool callback for learn (MentalModelInput) -> result.
-                  If None, learn tool is disabled.
+        expand_fn: Tool callback for expand (memory_ids, depth) -> result
         context: Optional additional context
         max_iterations: Maximum number of iterations before forcing response
         max_tokens: Maximum tokens for the final response
@@ -188,7 +220,6 @@ async def run_reflect_agent(
     Returns:
         ReflectAgentResult with final answer and metadata
     """
-    enable_learn = learn_fn is not None
     reflect_id = f"{bank_id[:8]}-{int(time.time() * 1000) % 100000}"
     start_time = time.time()
 
@@ -199,7 +230,7 @@ async def run_reflect_agent(
     directive_rules = _extract_directive_rules(directives) if directives else None
 
     # Get tools for this agent (with directive compliance field if directives exist)
-    tools = get_reflect_tools(enable_learn=enable_learn, directive_rules=directive_rules)
+    tools = get_reflect_tools(directive_rules=directive_rules)
 
     # Build initial messages (directives are injected into system prompt at START and END)
     system_prompt = build_system_prompt_for_tools(bank_profile, context, directives=directives)
@@ -209,7 +240,6 @@ async def run_reflect_agent(
     ]
 
     # Tracking
-    mental_models_created: list[str] = []
     total_tools_called = 0
     tool_trace: list[ToolCall] = []
     tool_trace_summary: list[dict[str, Any]] = []
@@ -218,45 +248,8 @@ async def run_reflect_agent(
 
     # Track available IDs for validation (prevents hallucinated citations)
     available_memory_ids: set[str] = set()
-    available_model_ids: set[str] = set()
-
-    # Pre-fetch mental models so the agent always starts with this knowledge
-    prefetch_start = time.time()
-    models_result = await lookup_fn(None)  # List all mental models
-    prefetch_duration = int((time.time() - prefetch_start) * 1000)
-
-    # Track available model IDs
-    if isinstance(models_result, dict) and "models" in models_result:
-        for model in models_result["models"]:
-            if "id" in model:
-                available_model_ids.add(model["id"])
-
-    # Add to context history for the agent
-    context_history.append({"tool": "list_mental_models", "output": models_result})
-
-    # Add to tool trace
-    tool_trace.append(
-        ToolCall(
-            tool="list_mental_models",
-            input={"tool": "list_mental_models"},
-            output=models_result,
-            duration_ms=prefetch_duration,
-            iteration=0,
-        )
-    )
-    tool_trace_summary.append(
-        {
-            "tool": "list_mental_models",
-            "input_summary": "(prefetch)",
-            "duration_ms": prefetch_duration,
-            "output_chars": len(json.dumps(models_result, default=str)),
-        }
-    )
-    total_tools_called += 1
-
-    # Include in the user message so the agent sees it
-    models_info = json.dumps(models_result, indent=2, default=str)
-    messages[1]["content"] = f"{query}\n\n## Available Mental Models (pre-fetched)\n```json\n{models_info}\n```"
+    available_reflection_ids: set[str] = set()
+    available_mental_model_ids: set[str] = set()
 
     def _get_llm_trace() -> list[LLMCall]:
         return [LLMCall(scope=c["scope"], duration_ms=c["duration_ms"]) for c in llm_trace]
@@ -315,7 +308,6 @@ async def run_reflect_agent(
                 structured_output=structured_output,
                 iterations=iteration + 1,
                 tools_called=total_tools_called,
-                mental_models_created=mental_models_created,
                 tool_trace=tool_trace,
                 llm_trace=_get_llm_trace(),
                 directives_applied=directives_applied,
@@ -334,12 +326,14 @@ async def run_reflect_agent(
             llm_duration = int((time.time() - llm_start) * 1000)
             llm_trace.append({"scope": f"agent_{iteration + 1}", "duration_ms": llm_duration})
 
-        except Exception:
-            llm_trace.append(
-                {"scope": f"agent_{iteration + 1}_err", "duration_ms": int((time.time() - llm_start) * 1000)}
-            )
+        except Exception as e:
+            err_duration = int((time.time() - llm_start) * 1000)
+            logger.warning(f"[REFLECT {reflect_id}] LLM error on iteration {iteration + 1}: {e} ({err_duration}ms)")
+            llm_trace.append({"scope": f"agent_{iteration + 1}_err", "duration_ms": err_duration})
             # Guardrail: If no evidence gathered yet, retry
-            has_gathered_evidence = bool(available_memory_ids) or bool(available_model_ids)
+            has_gathered_evidence = (
+                bool(available_memory_ids) or bool(available_reflection_ids) or bool(available_mental_model_ids)
+            )
             if not has_gathered_evidence and iteration < max_iterations - 1:
                 continue
             prompt = build_final_prompt(query, context_history, bank_profile, context)
@@ -366,7 +360,6 @@ async def run_reflect_agent(
                 structured_output=structured_output,
                 iterations=iteration + 1,
                 tools_called=total_tools_called,
-                mental_models_created=mental_models_created,
                 tool_trace=tool_trace,
                 llm_trace=_get_llm_trace(),
                 directives_applied=directives_applied,
@@ -390,7 +383,6 @@ async def run_reflect_agent(
                     structured_output=structured_output,
                     iterations=iteration + 1,
                     tools_called=total_tools_called,
-                    mental_models_created=mental_models_created,
                     tool_trace=tool_trace,
                     llm_trace=_get_llm_trace(),
                     directives_applied=directives_applied,
@@ -420,17 +412,18 @@ async def run_reflect_agent(
                 structured_output=structured_output,
                 iterations=iteration + 1,
                 tools_called=total_tools_called,
-                mental_models_created=mental_models_created,
                 tool_trace=tool_trace,
                 llm_trace=_get_llm_trace(),
                 directives_applied=directives_applied,
             )
 
-        # Check for done tool call (handle both 'done' and 'functions.done')
-        done_call = next((tc for tc in result.tool_calls if tc.name == "done" or tc.name == "functions.done"), None)
+        # Check for done tool call (handle various LLM output formats)
+        done_call = next((tc for tc in result.tool_calls if _is_done_tool(tc.name)), None)
         if done_call:
             # Guardrail: Require evidence before done
-            has_gathered_evidence = bool(available_memory_ids) or bool(available_model_ids)
+            has_gathered_evidence = (
+                bool(available_memory_ids) or bool(available_reflection_ids) or bool(available_mental_model_ids)
+            )
             if not has_gathered_evidence and iteration < max_iterations - 1:
                 # Add assistant message and fake tool result asking for evidence
                 messages.append(
@@ -443,9 +436,10 @@ async def run_reflect_agent(
                     {
                         "role": "tool",
                         "tool_call_id": done_call.id,
+                        "name": done_call.name,  # Required by Gemini
                         "content": json.dumps(
                             {
-                                "error": "You must call recall() or list_mental_models() to gather evidence before providing your final answer."
+                                "error": "You must search for information first. Use search_reflections(), search_mental_models(), or recall() before providing your final answer."
                             }
                         ),
                     }
@@ -456,10 +450,10 @@ async def run_reflect_agent(
             return await _process_done_tool(
                 done_call,
                 available_memory_ids,
-                available_model_ids,
+                available_reflection_ids,
+                available_mental_model_ids,
                 iteration + 1,
                 total_tools_called,
-                mental_models_created,
                 tool_trace,
                 _get_llm_trace(),
                 _log_completion,
@@ -469,8 +463,8 @@ async def run_reflect_agent(
                 response_schema=response_schema,
             )
 
-        # Execute other tools in parallel (exclude done and functions.done)
-        other_tools = [tc for tc in result.tool_calls if tc.name not in ("done", "functions.done")]
+        # Execute other tools in parallel (exclude done tool in all its format variants)
+        other_tools = [tc for tc in result.tool_calls if not _is_done_tool(tc.name)]
         if other_tools:
             # Add assistant message with tool calls
             messages.append(
@@ -482,7 +476,14 @@ async def run_reflect_agent(
 
             # Execute tools in parallel
             tool_tasks = [
-                _execute_tool_with_timing(tc, lookup_fn, recall_fn, expand_fn, learn_fn) for tc in other_tools
+                _execute_tool_with_timing(
+                    tc,
+                    search_reflections_fn,
+                    search_mental_models_fn,
+                    recall_fn,
+                    expand_fn,
+                )
+                for tc in other_tools
             ]
             tool_results = await asyncio.gather(*tool_tasks, return_exceptions=True)
             total_tools_called += len(other_tools)
@@ -490,43 +491,52 @@ async def run_reflect_agent(
             # Process results and add to messages
             for tc, result_data in zip(other_tools, tool_results):
                 if isinstance(result_data, Exception):
-                    # Tool execution failed - log and raise to fail the request
-                    logger.error(f"[REFLECT {reflect_id}] Tool {tc.name} failed with exception: {result_data}")
-                    raise RuntimeError(f"Reflect tool '{tc.name}' failed: {result_data}")
+                    # Tool execution failed - send error back to LLM so it can try again
+                    logger.warning(f"[REFLECT {reflect_id}] Tool {tc.name} failed with exception: {result_data}")
+                    output = {"error": f"Tool execution failed: {result_data}"}
+                    duration_ms = 0
+                else:
+                    output, duration_ms = result_data
 
-                output, duration_ms = result_data
+                # Normalize tool name for consistent tracking
+                normalized_tool_name = _normalize_tool_name(tc.name)
 
-                # Check if tool returned an error response
+                # Check if tool returned an error response - log but continue (LLM will see the error)
                 if isinstance(output, dict) and "error" in output:
-                    logger.error(f"[REFLECT {reflect_id}] Tool {tc.name} returned error: {output['error']}")
-                    raise RuntimeError(f"Reflect tool '{tc.name}' error: {output['error']}")
+                    logger.warning(
+                        f"[REFLECT {reflect_id}] Tool {normalized_tool_name} returned error: {output['error']}"
+                    )
 
-                # Track created mental models
-                if tc.name == "learn" and isinstance(output, dict) and "model_id" in output:
-                    mental_models_created.append(output["model_id"])
+                # Track available IDs from tool results (only for successful responses)
+                if (
+                    normalized_tool_name == "search_reflections"
+                    and isinstance(output, dict)
+                    and "reflections" in output
+                ):
+                    for reflection in output["reflections"]:
+                        if "id" in reflection:
+                            available_reflection_ids.add(reflection["id"])
 
-                # Track available memory IDs from recall
-                if tc.name == "recall" and isinstance(output, dict) and "memories" in output:
+                if (
+                    normalized_tool_name == "search_mental_models"
+                    and isinstance(output, dict)
+                    and "mental_models" in output
+                ):
+                    for mm in output["mental_models"]:
+                        if "id" in mm:
+                            available_mental_model_ids.add(mm["id"])
+
+                if normalized_tool_name == "recall" and isinstance(output, dict) and "memories" in output:
                     for memory in output["memories"]:
                         if "id" in memory:
                             available_memory_ids.add(memory["id"])
 
-                # Track available model IDs
-                if tc.name in ("list_mental_models", "get_mental_model") and isinstance(output, dict):
-                    if output.get("found") and "model" in output:
-                        model_id = output["model"].get("id")
-                        if model_id:
-                            available_model_ids.add(model_id)
-                    elif "models" in output:
-                        for model in output["models"]:
-                            if "id" in model:
-                                available_model_ids.add(model["id"])
-
                 # Add tool result message
                 messages.append(
                     {
                         "role": "tool",
                         "tool_call_id": tc.id,
+                        "name": tc.name,  # Required by Gemini
                         "content": json.dumps(output, default=str),
                     }
                 )
@@ -565,7 +575,6 @@ async def run_reflect_agent(
         text=answer,
         iterations=max_iterations,
         tools_called=total_tools_called,
-        mental_models_created=mental_models_created,
         tool_trace=tool_trace,
         llm_trace=_get_llm_trace(),
         directives_applied=directives_applied,
@@ -587,10 +596,10 @@ def _tool_call_to_dict(tc: "LLMToolCall") -> dict[str, Any]:
 async def _process_done_tool(
     done_call: "LLMToolCall",
     available_memory_ids: set[str],
-    available_model_ids: set[str],
+    available_reflection_ids: set[str],
+    available_mental_model_ids: set[str],
     iterations: int,
     total_tools_called: int,
-    mental_models_created: list[str],
     tool_trace: list[ToolCall],
     llm_trace: list[LLMCall],
     log_completion: Callable,
@@ -606,9 +615,10 @@ async def _process_done_tool(
     if not answer:
         answer = "No answer provided."
 
-    # Validate IDs
+    # Validate IDs (only include IDs that were actually retrieved)
     used_memory_ids = [mid for mid in args.get("memory_ids", []) if mid in available_memory_ids]
-    used_model_ids = [mid for mid in args.get("model_ids", []) if mid in available_model_ids]
+    used_reflection_ids = [rid for rid in args.get("reflection_ids", []) if rid in available_reflection_ids]
+    used_mental_model_ids = [mid for mid in args.get("mental_model_ids", []) if mid in available_mental_model_ids]
 
     # Generate structured output if schema provided
     structured_output = None
@@ -621,25 +631,32 @@ async def _process_done_tool(
         structured_output=structured_output,
         iterations=iterations,
         tools_called=total_tools_called,
-        mental_models_created=mental_models_created,
         tool_trace=tool_trace,
         llm_trace=llm_trace,
         used_memory_ids=used_memory_ids,
-        used_model_ids=used_model_ids,
+        used_reflection_ids=used_reflection_ids,
+        used_mental_model_ids=used_mental_model_ids,
         directives_applied=directives_applied,
     )
 
 
 async def _execute_tool_with_timing(
     tc: "LLMToolCall",
-    lookup_fn: Callable[[str | None], Awaitable[dict[str, Any]]],
+    search_reflections_fn: Callable[[str, int], Awaitable[dict[str, Any]]],
+    search_mental_models_fn: Callable[[str, int], Awaitable[dict[str, Any]]],
     recall_fn: Callable[[str, int], Awaitable[dict[str, Any]]],
     expand_fn: Callable[[list[str], str], Awaitable[dict[str, Any]]],
-    learn_fn: Callable[[MentalModelInput], Awaitable[dict[str, Any]]] | None = None,
 ) -> tuple[dict[str, Any], int]:
     """Execute a tool call and return result with timing."""
     start = time.time()
-    result = await _execute_tool(tc.name, tc.arguments, lookup_fn, recall_fn, expand_fn, learn_fn)
+    result = await _execute_tool(
+        tc.name,
+        tc.arguments,
+        search_reflections_fn,
+        search_mental_models_fn,
+        recall_fn,
+        expand_fn,
+    )
     duration_ms = int((time.time() - start) * 1000)
     return result, duration_ms
 
@@ -647,24 +664,28 @@ async def _execute_tool_with_timing(
 async def _execute_tool(
     tool_name: str,
     args: dict[str, Any],
-    lookup_fn: Callable[[str | None], Awaitable[dict[str, Any]]],
+    search_reflections_fn: Callable[[str, int], Awaitable[dict[str, Any]]],
+    search_mental_models_fn: Callable[[str, int], Awaitable[dict[str, Any]]],
     recall_fn: Callable[[str, int], Awaitable[dict[str, Any]]],
     expand_fn: Callable[[list[str], str], Awaitable[dict[str, Any]]],
-    learn_fn: Callable[[MentalModelInput], Awaitable[dict[str, Any]]] | None = None,
 ) -> dict[str, Any]:
     """Execute a single tool by name."""
-    # Normalize tool name - some LLMs return 'functions.done' instead of 'done'
-    if tool_name.startswith("functions."):
-        tool_name = tool_name[len("functions.") :]
+    # Normalize tool name for various LLM output formats
+    tool_name = _normalize_tool_name(tool_name)
 
-    if tool_name == "list_mental_models":
-        return await lookup_fn(None)
+    if tool_name == "search_reflections":
+        query = args.get("query")
+        if not query:
+            return {"error": "search_reflections requires a query parameter"}
+        max_results = args.get("max_results") or 5
+        return await search_reflections_fn(query, max_results)
 
-    elif tool_name == "get_mental_model":
-        model_id = args.get("model_id")
-        if not model_id:
-            return {"error": "get_mental_model requires model_id"}
-        return await lookup_fn(model_id)
+    elif tool_name == "search_mental_models":
+        query = args.get("query")
+        if not query:
+            return {"error": "search_mental_models requires a query parameter"}
+        max_tokens = max(args.get("max_tokens") or 5000, 1000)  # Default 5000, min 1000
+        return await search_mental_models_fn(query, max_tokens)
 
     elif tool_name == "recall":
         query = args.get("query")
@@ -673,15 +694,6 @@ async def _execute_tool(
         max_tokens = max(args.get("max_tokens") or 2048, 1000)  # Default 2048, min 1000
         return await recall_fn(query, max_tokens)
 
-    elif tool_name == "learn":
-        if learn_fn is None:
-            return {"error": "learn tool is not available"}
-        name = args.get("name")
-        description = args.get("description")
-        if not name or not description:
-            return {"error": "learn requires name and description"}
-        return await learn_fn(MentalModelInput(name=name, description=description))
-
     elif tool_name == "expand":
         memory_ids = args.get("memory_ids", [])
         if not memory_ids:
@@ -695,21 +707,22 @@ async def _execute_tool(
 
 def _summarize_input(tool_name: str, args: dict[str, Any]) -> str:
     """Create a summary of tool input for logging, showing all params."""
-    if tool_name == "list_mental_models":
-        return "()"
-    elif tool_name == "get_mental_model":
-        return f"(model_id={args.get('model_id', '?')})"
+    if tool_name == "search_reflections":
+        query = args.get("query", "")
+        query_preview = f"'{query[:30]}...'" if len(query) > 30 else f"'{query}'"
+        max_results = args.get("max_results") or 5
+        return f"(query={query_preview}, max_results={max_results})"
+    elif tool_name == "search_mental_models":
+        query = args.get("query", "")
+        query_preview = f"'{query[:30]}...'" if len(query) > 30 else f"'{query}'"
+        max_tokens = max(args.get("max_tokens") or 5000, 1000)
+        return f"(query={query_preview}, max_tokens={max_tokens})"
     elif tool_name == "recall":
         query = args.get("query", "")
         query_preview = f"'{query[:30]}...'" if len(query) > 30 else f"'{query}'"
         # Show actual value used (default 2048, min 1000)
         max_tokens = max(args.get("max_tokens") or 2048, 1000)
         return f"(query={query_preview}, max_tokens={max_tokens})"
-    elif tool_name == "learn":
-        name = args.get("name", "?")
-        desc = args.get("description", "")
-        desc_preview = f"'{desc[:20]}...'" if len(desc) > 20 else f"'{desc}'"
-        return f"(name='{name}', description={desc_preview})"
     elif tool_name == "expand":
         memory_ids = args.get("memory_ids", [])
         depth = args.get("depth", "chunk")
@@ -718,6 +731,9 @@ def _summarize_input(tool_name: str, args: dict[str, Any]) -> str:
         answer = args.get("answer", "")
         answer_preview = f"'{answer[:30]}...'" if len(answer) > 30 else f"'{answer}'"
         memory_ids = args.get("memory_ids", [])
-        model_ids = args.get("model_ids", [])
-        return f"(answer={answer_preview}, memory_ids={len(memory_ids)}, model_ids={len(model_ids)})"
+        reflection_ids = args.get("reflection_ids", [])
+        mental_model_ids = args.get("mental_model_ids", [])
+        return (
+            f"(answer={answer_preview}, mem={len(memory_ids)}, ref={len(reflection_ids)}, mm={len(mental_model_ids)})"
+        )
     return str(args)

hindsight-api/hindsight_api/engine/reflect/models.py

@@ -104,11 +104,15 @@ class ReflectAgentResult(BaseModel):
     )
     iterations: int = Field(default=0, description="Number of iterations taken")
     tools_called: int = Field(default=0, description="Total number of tool calls made")
-    mental_models_created: list[str] = Field(default_factory=list, description="IDs of mental models created/updated")
     tool_trace: list[ToolCall] = Field(default_factory=list, description="Trace of all tool calls made")
     llm_trace: list[LLMCall] = Field(default_factory=list, description="Trace of all LLM calls made")
     used_memory_ids: list[str] = Field(default_factory=list, description="Validated memory IDs actually used in answer")
-    used_model_ids: list[str] = Field(default_factory=list, description="Validated model IDs actually used in answer")
+    used_reflection_ids: list[str] = Field(
+        default_factory=list, description="Validated reflection IDs actually used in answer"
+    )
+    used_mental_model_ids: list[str] = Field(
+        default_factory=list, description="Validated mental model IDs actually used in answer"
+    )
     directives_applied: list[DirectiveInfo] = Field(
         default_factory=list, description="Directive mental models that affected this reflection"
     )

hindsight-api/hindsight_api/engine/reflect/observations.py

@@ -184,65 +184,3 @@ def compute_trend(
         return Trend.WEAKENING
     else:
         return Trend.STABLE
-
-
-class CandidateObservation(BaseModel):
-    """A candidate observation generated during the seed phase.
-
-    Candidates are preliminary observations that need evidence validation
-    before becoming full observations.
-    """
-
-    content: str = Field(description="The proposed observation content")
-    seed_memory_ids: list[str] = Field(default_factory=list, description="Memory IDs that inspired this candidate")
-
-
-class CandidateWithEvidence(BaseModel):
-    """A candidate observation with gathered supporting and contradicting evidence."""
-
-    candidate: CandidateObservation
-    supporting_memories: list[dict] = Field(default_factory=list, description="Memories that support this observation")
-    contradicting_memories: list[dict] = Field(
-        default_factory=list, description="Memories that contradict this observation"
-    )
-
-
-class MentalModelSnapshot(BaseModel):
-    """A versioned snapshot of a mental model's observations.
-
-    Used for tracking changes over time and enabling diff views.
-    """
-
-    version: int = Field(description="Version number (1-indexed)")
-    observations: list[Observation] = Field(default_factory=list, description="Observations at this version")
-    created_at: datetime = Field(
-        default_factory=lambda: datetime.now(timezone.utc), description="When this version was created"
-    )
-    reflect_summary: str | None = Field(default=None, description="Summary of changes in this version")
-
-
-def verify_evidence_quotes(
-    observation: Observation,
-    memories: dict[str, str],
-) -> tuple[bool, list[str]]:
-    """Verify that all evidence quotes exist in the referenced memories.
-
-    Args:
-        observation: The observation to verify
-        memories: Dict mapping memory_id to memory content
-
-    Returns:
-        Tuple of (is_valid, list of error messages)
-    """
-    errors = []
-
-    for evidence in observation.evidence:
-        memory_content = memories.get(evidence.memory_id)
-        if memory_content is None:
-            errors.append(f"Memory {evidence.memory_id} not found")
-            continue
-
-        if evidence.quote not in memory_content:
-            errors.append(f"Quote not found in memory {evidence.memory_id}: '{evidence.quote[:50]}...'")
-
-    return len(errors) == 0, errors

hindsight-api/hindsight_api/engine/reflect/prompts.py

@@ -1,5 +1,10 @@
 """
 System prompts for the reflect agent.
+
+The reflect agent uses hierarchical retrieval:
+1. search_reflections - User-curated summaries (highest quality)
+2. search_mental_models - Consolidated knowledge with freshness awareness
+3. recall - Raw facts as ground truth fallback
 """
 
 import json
@@ -11,7 +16,7 @@ def _extract_directive_rules(directives: list[dict[str, Any]]) -> list[str]:
     Extract directive rules as a list of strings.
 
     Args:
-        directives: List of directive mental models with observations
+        directives: List of directives with name and content
 
     Returns:
         List of directive rule strings
@@ -19,25 +24,34 @@ def _extract_directive_rules(directives: list[dict[str, Any]]) -> list[str]:
     rules = []
     for directive in directives:
         directive_name = directive.get("name", "")
-        observations = directive.get("observations", [])
-        if observations:
-            for obs in observations:
-                # Support both Pydantic Observation objects and dicts
-                if hasattr(obs, "title"):
-                    title = obs.title
-                    content = obs.content
-                else:
-                    title = obs.get("title", "")
-                    content = obs.get("content", "")
-                if title and content:
-                    rules.append(f"**{title}**: {content}")
-                elif content:
-                    rules.append(content)
-        elif directive_name:
-            # Fallback to description if no observations
-            desc = directive.get("description", "")
-            if desc:
-                rules.append(f"**{directive_name}**: {desc}")
+        # New format: directives have direct content field
+        content = directive.get("content", "")
+        if content:
+            if directive_name:
+                rules.append(f"**{directive_name}**: {content}")
+            else:
+                rules.append(content)
+        else:
+            # Legacy format: check for observations
+            observations = directive.get("observations", [])
+            if observations:
+                for obs in observations:
+                    # Support both Pydantic Observation objects and dicts
+                    if hasattr(obs, "title"):
+                        title = obs.title
+                        obs_content = obs.content
+                    else:
+                        title = obs.get("title", "")
+                        obs_content = obs.get("content", "")
+                    if title and obs_content:
+                        rules.append(f"**{title}**: {obs_content}")
+                    elif obs_content:
+                        rules.append(obs_content)
+            elif directive_name:
+                # Fallback to description
+                desc = directive.get("description", "")
+                if desc:
+                    rules.append(f"**{directive_name}**: {desc}")
     return rules
 
 
@@ -111,24 +125,25 @@ def build_system_prompt_for_tools(
     bank_profile: dict[str, Any],
     context: str | None = None,
     directives: list[dict[str, Any]] | None = None,
+    has_reflections: bool = False,
 ) -> str:
     """
     Build the system prompt for tool-calling reflect agent.
 
-    This is a simplified prompt since tools are defined separately via the tools parameter.
+    The agent uses hierarchical retrieval:
+    1. search_reflections - User-curated summaries (try first, if available)
+    2. search_mental_models - Consolidated knowledge with freshness
+    3. recall - Raw facts as ground truth
 
     Args:
         bank_profile: Bank profile with name and mission
         context: Optional additional context
         directives: Optional list of directive mental models to inject as hard rules
+        has_reflections: Whether the bank has any reflections (skip if not)
     """
     name = bank_profile.get("name", "Assistant")
     mission = bank_profile.get("mission", "")
 
-    no_info_rule = (
-        "- Only say 'I don't have information' AFTER trying list_mental_models AND recall with no relevant results"
-    )
-
     parts = []
 
     # Inject directives at the VERY START for maximum prominence
@@ -147,8 +162,7 @@ def build_system_prompt_for_tools(
             "## CRITICAL RULES",
             "- You must NEVER fabricate information that has no basis in retrieved data",
             "- You SHOULD synthesize, infer, and reason from the retrieved memories",
-            "- You MUST call recall() before saying you don't have information",
-            no_info_rule,
+            "- You MUST search before saying you don't have information",
             "",
             "## How to Reason",
             "- If memories mention someone did an activity, you can infer they likely enjoyed it",
@@ -156,7 +170,56 @@ def build_system_prompt_for_tools(
             "- Be a thoughtful interpreter, not just a literal repeater",
             "- When the exact answer isn't stated, use what IS stated to give the best answer",
             "",
-            "## Query Strategy (IMPORTANT)",
+            "## HIERARCHICAL RETRIEVAL STRATEGY",
+            "",
+        ]
+    )
+
+    # Build retrieval levels based on what's available
+    if has_reflections:
+        parts.extend(
+            [
+                "You have access to THREE levels of knowledge. Use them in this order:",
+                "",
+                "### 1. REFLECTIONS (search_reflections) - Try First",
+                "- User-curated summaries about specific topics",
+                "- HIGHEST quality - manually created and maintained",
+                "- If a relevant reflection exists and is FRESH, it may fully answer the question",
+                "- Check `is_stale` field - if stale, also verify with lower levels",
+                "",
+                "### 2. MENTAL MODELS (search_mental_models) - Second Priority",
+                "- Auto-consolidated knowledge from memories",
+                "- Check `is_stale` field - if stale, ALSO use recall() to verify",
+                "- Good for understanding patterns and summaries",
+                "",
+                "### 3. RAW FACTS (recall) - Ground Truth",
+                "- Individual memories (world facts and experiences)",
+                "- Use when: no reflections/models exist, they're stale, or you need specific details",
+                "- This is the source of truth that other levels are built from",
+                "",
+            ]
+        )
+    else:
+        parts.extend(
+            [
+                "You have access to TWO levels of knowledge. Use them in this order:",
+                "",
+                "### 1. MENTAL MODELS (search_mental_models) - Try First",
+                "- Auto-consolidated knowledge from memories",
+                "- Check `is_stale` field - if stale, ALSO use recall() to verify",
+                "- Good for understanding patterns and summaries",
+                "",
+                "### 2. RAW FACTS (recall) - Ground Truth",
+                "- Individual memories (world facts and experiences)",
+                "- Use when: no mental models exist, they're stale, or you need specific details",
+                "- This is the source of truth that mental models are built from",
+                "",
+            ]
+        )
+
+    parts.extend(
+        [
+            "## Query Strategy",
             "recall() uses semantic search. NEVER just echo the user's question - decompose it into targeted searches:",
             "",
             "BAD: User asks 'recurring lesson themes between students' → recall('recurring lesson themes between students')",
@@ -164,44 +227,41 @@ def build_system_prompt_for_tools(
             "  1. recall('lessons') - find all lesson-related memories",
             "  2. recall('teaching sessions') - alternative phrasing",
             "  3. recall('student progress') - find student-related memories",
-            "  4. recall('topics taught') - find subject matter",
             "",
             "Think: What ENTITIES and CONCEPTS does this question involve? Search for each separately.",
-            "- Questions about patterns → search for the individual instances first",
-            "- Questions comparing things → search for each thing separately",
-            "- Questions about relationships → search for each party involved",
             "",
             "## Workflow",
         ]
     )
 
-    # Answer mode: include mental model lookup in workflow
+    if has_reflections:
+        parts.extend(
+            [
+                "1. First, try search_reflections() - check if a curated summary exists",
+                "2. If no reflection or it's stale, try search_mental_models() for consolidated knowledge",
+                "3. If mental models are stale OR you need specific details, use recall() for raw facts",
+                "4. Use expand() if you need more context on specific memories",
+                "5. When ready, call done() with your answer and supporting IDs",
+            ]
+        )
+    else:
+        parts.extend(
+            [
+                "1. First, try search_mental_models() - check for consolidated knowledge",
+                "2. If mental models are stale OR you need specific details, use recall() for raw facts",
+                "3. Use expand() if you need more context on specific memories",
+                "4. When ready, call done() with your answer and supporting IDs",
+            ]
+        )
+
     parts.extend(
         [
-            "1. Review the pre-fetched mental models for relevant synthesized knowledge",
-            "2. If relevant, call get_mental_model(model_id) for full observations",
-            "3. DECOMPOSE the question into component searches (see Query Strategy above)",
-            "   - Identify entities and concepts in the question",
-            "   - Search for each separately with targeted queries",
-            "4. Run multiple recall() calls - don't just echo the user's question",
-            "5. Use expand() if you need more context on specific memories",
-            "6. BEFORE answering: Check if any person/project/concept from the memories deserves a mental model - use learn() if so",
-            "7. When ready, call done() with your answer and supporting memory_ids",
-            "",
-            "## When to Use learn() - IMPORTANT",
-            "ACTIVELY look for opportunities to use learn() when you discover:",
-            "- A person mentioned in 2+ memories who has no mental model yet",
-            "- A project or concept the user asks about that has no mental model",
-            "- A pattern or topic worth tracking for future questions",
-            "",
-            "DO NOT wait to be asked - proactively create models when you see the need.",
-            "Example: learn(name='Project Alpha', description='Track goals, status, and key decisions for Project Alpha')",
             "",
             "## Output Format: Plain Text Answer",
             "Call done() with a plain text 'answer' field.",
             "- Do NOT use markdown formatting",
             "- NEVER include memory IDs, UUIDs, or 'Memory references' in the answer text",
-            "- Put memory IDs ONLY in the memory_ids array parameter, not in the answer",
+            "- Put IDs ONLY in the memory_ids/reflection_ids/mental_model_ids arrays, not in the answer",
         ]
     )
 
@@ -295,9 +355,10 @@ def build_agent_prompt(
     else:
         parts.append(
             "\n## Instructions\n"
-            "Start by calling list_mental_models() to see available mental models - they contain pre-synthesized knowledge. "
-            "If a relevant model exists, use get_mental_model(model_id) to get its observations. "
-            "Then use recall(query) for specific details not covered by mental models."
+            "Start by searching for relevant information using the hierarchical retrieval strategy:\n"
+            "1. Try search_reflections() first for curated summaries\n"
+            "2. Try search_mental_models() for consolidated knowledge\n"
+            "3. Use recall() for specific details or to verify stale data"
         )
 
     return "\n".join(parts)
@@ -377,386 +438,3 @@ Your approach:
 
 Only say "I don't have information" if the retrieved data is truly unrelated to the question.
 Do NOT fabricate information that has no basis in the retrieved data."""
-
-
-# =============================================================================
-# 4-Phase Mental Model Reflect Prompts
-# =============================================================================
-
-SEED_PHASE_SYSTEM_PROMPT = """You are analyzing memories to discover NEW patterns and generate candidate observations.
-
-Your task is to identify potential observations (beliefs, preferences, patterns, behaviors) that could be part of a mental model about this person/topic.
-
-## Important: Avoid Redundancy
-If existing observations are provided, DO NOT generate candidates that are essentially the same.
-Focus on discovering NEW patterns not already covered by existing observations.
-
-## Rules
-- Generate 5-15 candidate observations for NEW patterns only
-- Each candidate should be specific and testable (can be supported or contradicted by evidence)
-- Note which memory IDs inspired each candidate (these are seeds, not final evidence)
-- Focus on patterns that appear MULTIPLE TIMES across many memories - the more the better
-- The best candidates are ones you can find 10, 20, or even 50+ supporting memories for
-- Skip patterns that are already covered by existing observations
-
-## Output Format
-Return a JSON array of candidate observations:
-```json
-{
-  "candidates": [
-    {
-      "content": "The specific observation/belief/pattern - be detailed and specific",
-      "seed_memory_ids": ["memory_id_1", "memory_id_2", "memory_id_3"]
-    }
-  ]
-}
-```
-
-Focus on patterns that appear multiple times or have strong signals. Don't generate obvious or trivial observations.
-Prefer candidates with MORE seed memories - they're more likely to be real patterns.
-Return an empty candidates array if no genuinely new patterns are found."""
-
-
-def build_seed_phase_prompt(
-    memories: list[dict],
-    topic: str | None = None,
-    existing_observations: list[dict] | None = None,
-) -> str:
-    """Build the user prompt for the seed phase.
-
-    Args:
-        memories: List of memories to analyze
-        topic: Optional topic focus for the mental model
-        existing_observations: Optional list of existing observations to avoid rediscovering
-    """
-    parts = []
-
-    if topic:
-        parts.append(f"## Topic Focus\n{topic}\n")
-
-    # Include existing observations so we don't rediscover them
-    if existing_observations:
-        parts.append("## Existing Observations (DO NOT regenerate these)")
-        parts.append("These patterns are already tracked. Focus on discovering NEW patterns:\n")
-        for i, obs in enumerate(existing_observations, 1):
-            title = obs.get("title", "")
-            content = obs.get("content", "")
-            parts.append(f"{i}. **{title}**: {content}\n")
-        parts.append("")
-
-    parts.append("## Memories to Analyze")
-    parts.append("Review these memories and identify patterns, preferences, beliefs, and behaviors:\n")
-
-    for mem in memories:
-        mem_id = mem.get("id", "unknown")
-        content = mem.get("content", mem.get("text", ""))
-        timestamp = mem.get("timestamp", mem.get("created_at", ""))
-        parts.append(f"[{mem_id}] ({timestamp}): {content}\n")
-
-    parts.append("\n## Instructions")
-    if existing_observations:
-        parts.append("Generate candidate observations for NEW patterns not already covered above.")
-        parts.append("If all patterns are already covered by existing observations, return an empty candidates array.")
-    else:
-        parts.append("Generate candidate observations based on patterns you see in these memories.")
-    parts.append("Look for: recurring themes, stated preferences, behavioral patterns, beliefs, values, goals.")
-
-    return "\n".join(parts)
-
-
-VALIDATE_PHASE_SYSTEM_PROMPT = """You are validating candidate observations against evidence.
-
-For each candidate, you have:
-- Supporting memories (evidence FOR the observation)
-- Contradicting memories (evidence AGAINST the observation)
-
-## Your Task
-1. Evaluate each candidate based on the evidence
-2. For valid candidates, extract EXACT QUOTES from supporting memories
-3. Discard candidates with insufficient or contradicting evidence
-4. Merge similar candidates into single, refined observations
-
-## Rules for Quotes
-- Quotes must be EXACT text from the memory, not paraphrased
-- Each quote should directly support the observation
-- The MORE evidence quotes, the BETTER - don't limit yourself, include ALL relevant quotes (10, 20, 50+)
-- Observations with only 1-2 quotes are weak and should be discarded unless the evidence is exceptionally strong
-- Stronger observations have more supporting evidence - aim for comprehensive coverage
-
-## Output Format
-Return validated observations with evidence:
-```json
-{
-  "observations": [
-    {
-      "title": "Short descriptive title (3-8 words) - like a headline",
-      "content": "The full observation content - detailed explanation of the pattern/belief",
-      "evidence": [
-        {
-          "memory_id": "exact_memory_id",
-          "quote": "Exact quote from the memory text",
-          "relevance": "Brief explanation of how this supports the observation",
-          "timestamp": "2024-01-15T10:00:00Z"
-        }
-      ]
-    }
-  ],
-  "discarded": [
-    {
-      "content": "The discarded candidate",
-      "reason": "Why it was discarded (insufficient evidence, contradicted, etc.)"
-    }
-  ],
-  "merged": [
-    {
-      "from": ["candidate 1 content", "candidate 2 content"],
-      "into": "The merged observation content"
-    }
-  ]
-}
-```
-
-## Title Guidelines
-- Title should be a SHORT label (like "Prefers morning meetings" or "Coffee enthusiast")
-- NOT a truncated version of the content
-- Think of it as a category/tag for the observation
-
-Be rigorous: only keep observations with clear, verifiable evidence from multiple memories."""
-
-
-def build_validate_phase_prompt(candidates_with_evidence: list[dict]) -> str:
-    """Build the user prompt for the validate phase."""
-    parts = ["## Candidates to Validate\n"]
-
-    for i, item in enumerate(candidates_with_evidence, 1):
-        candidate = item.get("candidate", {})
-        supporting = item.get("supporting_memories", [])
-        contradicting = item.get("contradicting_memories", [])
-
-        parts.append(f"### Candidate {i}: {candidate.get('content', '')}")
-
-        if supporting:
-            parts.append("\n**Supporting Evidence:**")
-            for mem in supporting:
-                mem_id = mem.get("id", "unknown")
-                content = mem.get("content", mem.get("text", ""))
-                timestamp = mem.get("timestamp", mem.get("created_at", ""))
-                parts.append(f"- [{mem_id}] ({timestamp}): {content}")
-
-        if contradicting:
-            parts.append("\n**Contradicting Evidence:**")
-            for mem in contradicting:
-                mem_id = mem.get("id", "unknown")
-                content = mem.get("content", mem.get("text", ""))
-                timestamp = mem.get("timestamp", mem.get("created_at", ""))
-                parts.append(f"- [{mem_id}] ({timestamp}): {content}")
-
-        if not supporting and not contradicting:
-            parts.append("\n*No additional evidence found*")
-
-        parts.append("")
-
-    parts.append("## Instructions")
-    parts.append("1. Evaluate each candidate based on its evidence")
-    parts.append("2. Keep candidates with strong supporting evidence")
-    parts.append("3. Discard candidates with no evidence or strong contradictions")
-    parts.append("4. Merge similar candidates")
-    parts.append("5. Extract EXACT quotes (copy-paste from memory text) for evidence")
-
-    return "\n".join(parts)
-
-
-COMPARE_PHASE_SYSTEM_PROMPT = """You are merging new observations with an existing mental model.
-
-You have:
-- EXISTING observations (from the current mental model)
-- NEW observations (from this reflect cycle)
-
-## Your Task
-Produce the final, complete mental model by:
-1. Keeping existing observations that are still valid
-2. Updating existing observations with new evidence (ADD new evidence to existing)
-3. Adding new observations that don't overlap with existing
-4. Removing existing observations that are contradicted by new evidence
-5. Merging overlapping observations
-
-## Rules
-- The final model should have no contradictions
-- Each observation must have evidence with exact quotes
-- COMBINE evidence from both existing and new observations
-- If an existing observation has new supporting evidence, ADD ALL the new evidence to it
-- Include ALL relevant evidence - the more quotes the better (10, 20, 50+ is great)
-- Observations with more evidence are more reliable - don't limit the number of quotes
-
-## Output Format
-Return the complete, final mental model:
-```json
-{
-  "observations": [
-    {
-      "title": "Short descriptive title (3-8 words)",
-      "content": "Full observation content - detailed explanation",
-      "evidence": [
-        {
-          "memory_id": "id",
-          "quote": "exact quote",
-          "relevance": "explanation",
-          "timestamp": "ISO timestamp"
-        }
-      ],
-      "created_at": "ISO timestamp of when observation was first created"
-    }
-  ],
-  "changes": {
-    "kept": ["Observation that was kept unchanged"],
-    "updated": [{"from": "old content", "to": "new content", "reason": "why"}],
-    "added": ["New observation that was added"],
-    "removed": [{"content": "removed observation", "reason": "why removed"}],
-    "merged": [{"from": ["obs1", "obs2"], "into": "merged observation"}]
-  }
-}
-```"""
-
-
-def build_compare_phase_prompt(
-    existing_observations: list[dict],
-    new_observations: list[dict],
-) -> str:
-    """Build the user prompt for the compare phase."""
-    parts = []
-
-    parts.append("## Existing Mental Model Observations")
-    if existing_observations:
-        for i, obs in enumerate(existing_observations, 1):
-            title = obs.get("title", "")
-            content = obs.get("content", obs.get("text", ""))
-            evidence = obs.get("evidence", [])
-            parts.append(f"\n### Existing {i}: {title}")
-            parts.append(f"Content: {content}")
-            if evidence:
-                parts.append(f"Evidence ({len(evidence)} items):")
-                for ev in evidence[:5]:  # Show max 5 evidence items
-                    parts.append(f'  - [{ev.get("memory_id", "?")}]: "{ev.get("quote", "")}"')
-                if len(evidence) > 5:
-                    parts.append(f"  ... and {len(evidence) - 5} more")
-    else:
-        parts.append("*No existing observations*")
-
-    parts.append("\n## New Observations from This Reflect")
-    if new_observations:
-        for i, obs in enumerate(new_observations, 1):
-            title = obs.get("title", "")
-            content = obs.get("content", "")
-            evidence = obs.get("evidence", [])
-            parts.append(f"\n### New {i}: {title}")
-            parts.append(f"Content: {content}")
-            if evidence:
-                parts.append(f"Evidence ({len(evidence)} items):")
-                for ev in evidence:
-                    parts.append(f'  - [{ev.get("memory_id", "?")}]: "{ev.get("quote", "")}"')
-    else:
-        parts.append("*No new observations*")
-
-    parts.append("\n## Instructions")
-    parts.append("Merge these into a coherent, non-contradictory mental model.")
-    parts.append("Preserve all valid evidence. Remove stale or contradicted observations.")
-
-    return "\n".join(parts)
-
-
-# =============================================================================
-# UPDATE EXISTING Phase Prompts (for diff-based refresh)
-# =============================================================================
-
-UPDATE_EXISTING_SYSTEM_PROMPT = """You are updating existing observations with newly found evidence.
-
-For each existing observation, you have been given:
-- The original observation (title, content, existing evidence)
-- Newly found supporting memories
-- Newly found contradicting memories
-
-## Your Task
-1. Extract EXACT QUOTES from new supporting memories to add to the observation
-2. Flag observations with strong contradicting evidence for potential removal
-3. Keep existing evidence intact - only ADD new evidence
-
-## Rules for Quotes
-- Quotes must be EXACT text from the memory, not paraphrased
-- Each quote should directly support the observation
-- Include ALL relevant quotes from the new memories
-
-## Output Format
-Return updated observations with new evidence:
-```json
-{
-  "updated_observations": [
-    {
-      "title": "Original title",
-      "content": "Original content",
-      "existing_evidence_count": 5,
-      "new_evidence": [
-        {
-          "memory_id": "exact_memory_id",
-          "quote": "Exact quote from the memory text",
-          "relevance": "Brief explanation of how this supports the observation",
-          "timestamp": "2024-01-15T10:00:00Z"
-        }
-      ],
-      "has_contradiction": false,
-      "contradiction_note": null
-    }
-  ]
-}
-```
-
-If an observation has strong contradicting evidence, set has_contradiction=true and explain in contradiction_note."""
-
-
-def build_update_existing_prompt(observations_with_evidence: list[dict]) -> str:
-    """Build the user prompt for the update existing phase.
-
-    Args:
-        observations_with_evidence: List of existing observations with new evidence found
-    """
-    parts = ["## Existing Observations to Update\n"]
-
-    for i, item in enumerate(observations_with_evidence, 1):
-        obs = item.get("observation", {})
-        supporting = item.get("supporting_memories", [])
-        contradicting = item.get("contradicting_memories", [])
-
-        title = obs.get("title", "")
-        content = obs.get("content", "")
-        existing_evidence = obs.get("evidence", [])
-
-        parts.append(f"### Observation {i}: {title}")
-        parts.append(f"Content: {content}")
-        parts.append(f"Existing evidence count: {len(existing_evidence)}")
-
-        if supporting:
-            parts.append("\n**New Supporting Memories:**")
-            for mem in supporting:
-                mem_id = mem.get("id", "unknown")
-                mem_content = mem.get("content", mem.get("text", ""))
-                timestamp = mem.get("timestamp", mem.get("created_at", ""))
-                parts.append(f"- [{mem_id}] ({timestamp}): {mem_content}")
-
-        if contradicting:
-            parts.append("\n**New Contradicting Memories:**")
-            for mem in contradicting:
-                mem_id = mem.get("id", "unknown")
-                mem_content = mem.get("content", mem.get("text", ""))
-                timestamp = mem.get("timestamp", mem.get("created_at", ""))
-                parts.append(f"- [{mem_id}] ({timestamp}): {mem_content}")
-
-        if not supporting and not contradicting:
-            parts.append("\n*No new evidence found*")
-
-        parts.append("")
-
-    parts.append("## Instructions")
-    parts.append("1. Extract EXACT quotes from new supporting memories")
-    parts.append("2. Flag observations with strong contradictions")
-    parts.append("3. Return the updated observations with new evidence added")
-
-    return "\n".join(parts)

hindsight-api/hindsight_api/engine/reflect/tools.py

@@ -1,16 +1,17 @@
 """
 Tool implementations for the reflect agent.
+
+Implements hierarchical retrieval:
+1. search_reflections - User-curated summaries (highest quality)
+2. search_mental_models - Consolidated knowledge with freshness
+3. recall - Raw facts as ground truth
 """
 
 import logging
-import re
 import uuid
-from datetime import datetime, timezone
+from datetime import datetime, timedelta, timezone
 from typing import TYPE_CHECKING, Any
 
-from .models import MentalModelInput
-from .observations import Observation, ObservationEvidence, Trend
-
 if TYPE_CHECKING:
     from asyncpg import Connection
 
@@ -19,156 +20,216 @@ if TYPE_CHECKING:
 
 logger = logging.getLogger(__name__)
 
-
-def generate_model_id(name: str) -> str:
-    """Generate a stable ID from mental model name."""
-    # Normalize: lowercase, replace spaces/special chars with hyphens
-    normalized = re.sub(r"[^a-z0-9]+", "-", name.lower()).strip("-")
-    # Truncate to reasonable length
-    return normalized[:50]
+# Mental model is considered stale if not updated in this many days
+STALE_THRESHOLD_DAYS = 7
 
 
-def _parse_observations(observations_raw: list) -> list[Observation]:
-    """Parse raw observation dicts into typed Observation models."""
-    observations: list[Observation] = []
-    for obs in observations_raw:
-        if not isinstance(obs, dict):
-            continue
-
-        try:
-            parsed = Observation(
-                title=obs.get("title", ""),
-                content=obs.get("content", ""),
-                evidence=[
-                    ObservationEvidence(
-                        memory_id=ev.get("memory_id", ""),
-                        quote=ev.get("quote", ""),
-                        relevance=ev.get("relevance", ""),
-                        timestamp=ev.get("timestamp"),
-                    )
-                    for ev in obs.get("evidence", [])
-                    if isinstance(ev, dict)
-                ],
-                created_at=obs.get("created_at"),
-            )
-            observations.append(parsed)
-        except Exception as e:
-            logger.warning(f"Failed to parse observation: {e}")
-            continue
-
-    return observations
-
-
-async def tool_lookup(
+async def tool_search_reflections(
     conn: "Connection",
     bank_id: str,
-    model_id: str | None = None,
+    query: str,
+    query_embedding: list[float],
+    max_results: int = 5,
     tags: list[str] | None = None,
     tags_match: str = "any",
+    exclude_ids: list[str] | None = None,
 ) -> dict[str, Any]:
     """
-    List or get mental models.
+    Search user-curated reflections by semantic similarity.
+
+    Reflections are high-quality, manually created summaries about specific topics.
+    They should be searched FIRST as they represent the most reliable synthesized knowledge.
 
     Args:
         conn: Database connection
         bank_id: Bank identifier
-        model_id: Optional specific model ID to get (if None, lists all)
-        tags: Optional tags to filter models (when listing)
+        query: Search query (for logging/tracing)
+        query_embedding: Pre-computed embedding for semantic search
+        max_results: Maximum number of reflections to return
+        tags: Optional tags to filter reflections
         tags_match: How to match tags - "any" (OR), "all" (AND)
+        exclude_ids: Optional list of reflection IDs to exclude (e.g., when refreshing a reflection)
 
     Returns:
-        Dict with either a list of models or a single model's details
+        Dict with matching reflections including content and freshness info
     """
-    if model_id:
-        # Get specific mental model with full details including observations
-        row = await conn.fetchrow(
-            """
-            SELECT id, subtype, name, description, observations, entity_id, last_updated
-            FROM mental_models
-            WHERE id = $1 AND bank_id = $2
-            """,
-            model_id,
-            bank_id,
-        )
-        if row:
-            # Parse observations JSON
-            obs_data = row["observations"] or {"observations": []}
-            if isinstance(obs_data, str):
-                import json
+    from ..memory_engine import fq_table
 
-                obs_data = json.loads(obs_data)
-            observations_raw = obs_data.get("observations", []) if isinstance(obs_data, dict) else obs_data
+    # Build filters dynamically
+    filters = ""
+    params: list[Any] = [bank_id, str(query_embedding), max_results]
+    next_param = 4
 
-            # Parse observations into typed models
-            observations = _parse_observations(observations_raw)
-
-            return {
-                "found": True,
-                "model": {
-                    "id": row["id"],
-                    "subtype": row["subtype"],
-                    "name": row["name"],
-                    "description": row["description"],
-                    "observations": observations,
-                    "entity_id": str(row["entity_id"]) if row["entity_id"] else None,
-                    "last_updated": row["last_updated"].isoformat() if row["last_updated"] else None,
-                },
-            }
-        return {"found": False, "model_id": model_id}
-    else:
-        # List mental models (compact: id, name, description only)
-        # Full observations are retrieved via get_mental_model(model_id)
-        # NOTE: Directives (subtype='directive') are excluded from listing -
-        # they are injected into the system prompt, not discoverable via tools
-        # Filter by tags if provided
-        if tags:
-            if tags_match == "all":
-                # All tags must match
-                rows = await conn.fetch(
-                    """
-                    SELECT id, subtype, name, description
-                    FROM mental_models
-                    WHERE bank_id = $1 AND tags @> $2::varchar[] AND subtype != 'directive'
-                    ORDER BY last_updated DESC NULLS LAST, created_at DESC
-                    """,
-                    bank_id,
-                    tags,
-                )
-            else:
-                # Any tag matches (OR) - default
-                rows = await conn.fetch(
-                    """
-                    SELECT id, subtype, name, description
-                    FROM mental_models
-                    WHERE bank_id = $1 AND tags && $2::varchar[] AND subtype != 'directive'
-                    ORDER BY last_updated DESC NULLS LAST, created_at DESC
-                    """,
-                    bank_id,
-                    tags,
-                )
+    if tags:
+        if tags_match == "all":
+            filters += f" AND tags @> ${next_param}::varchar[]"
         else:
-            rows = await conn.fetch(
-                """
-                SELECT id, subtype, name, description
-                FROM mental_models
-                WHERE bank_id = $1 AND subtype != 'directive'
-                ORDER BY last_updated DESC NULLS LAST, created_at DESC
+            filters += f" AND (tags && ${next_param}::varchar[] OR tags IS NULL OR tags = '{{}}')"
+        params.append(tags)
+        next_param += 1
+
+    if exclude_ids:
+        filters += f" AND id != ALL(${next_param}::uuid[])"
+        params.append(exclude_ids)
+        next_param += 1
+
+    # Search reflections by embedding similarity
+    rows = await conn.fetch(
+        f"""
+        SELECT
+            id, name, content, reflect_response,
+            tags, created_at, last_refreshed_at,
+            1 - (embedding <=> $2::vector) as relevance
+        FROM {fq_table("reflections")}
+        WHERE bank_id = $1 AND embedding IS NOT NULL {filters}
+        ORDER BY embedding <=> $2::vector
+        LIMIT $3
+        """,
+        *params,
+    )
+
+    now = datetime.now(timezone.utc)
+    reflections = []
+
+    for row in rows:
+        last_refreshed_at = row["last_refreshed_at"]
+        if last_refreshed_at and last_refreshed_at.tzinfo is None:
+            last_refreshed_at = last_refreshed_at.replace(tzinfo=timezone.utc)
+
+        # Calculate freshness
+        is_stale = False
+        if last_refreshed_at:
+            age = now - last_refreshed_at
+            is_stale = age > timedelta(days=STALE_THRESHOLD_DAYS)
+
+        reflections.append(
+            {
+                "id": str(row["id"]),
+                "name": row["name"],
+                "content": row["content"],
+                "reflect_response": row["reflect_response"],
+                "tags": row["tags"] or [],
+                "relevance": round(row["relevance"], 4),
+                "updated_at": last_refreshed_at.isoformat() if last_refreshed_at else None,
+                "is_stale": is_stale,
+            }
+        )
+
+    return {
+        "query": query,
+        "count": len(reflections),
+        "reflections": reflections,
+    }
+
+
+async def tool_search_mental_models(
+    memory_engine: "MemoryEngine",
+    bank_id: str,
+    query: str,
+    request_context: "RequestContext",
+    max_tokens: int = 5000,
+    tags: list[str] | None = None,
+    tags_match: str = "any",
+    last_consolidated_at: datetime | None = None,
+    pending_consolidation: int = 0,
+) -> dict[str, Any]:
+    """
+    Search consolidated mental models using recall with include_mental_models.
+
+    Mental models are auto-generated from memories. Returns freshness info
+    so the agent knows if it should also verify with recall().
+
+    Args:
+        memory_engine: Memory engine instance
+        bank_id: Bank identifier
+        query: Search query
+        request_context: Request context for authentication
+        max_tokens: Maximum tokens for results (default 5000)
+        tags: Optional tags to filter models
+        tags_match: How to match tags - "any" (OR), "all" (AND)
+        last_consolidated_at: When consolidation last ran (for staleness check)
+        pending_consolidation: Number of memories waiting to be consolidated
+
+    Returns:
+        Dict with matching mental models including freshness info
+    """
+    from ..memory_engine import fq_table
+
+    # Use recall to search mental models (they come back in results field when fact_type=["mental_model"])
+    result = await memory_engine.recall_async(
+        bank_id=bank_id,
+        query=query,
+        fact_type=["mental_model"],  # Only retrieve mental models
+        max_tokens=max_tokens,  # Token budget controls how many mental models are returned
+        enable_trace=False,
+        request_context=request_context,
+        tags=tags,
+        tags_match=tags_match,
+        _connection_budget=1,
+        _quiet=True,
+    )
+
+    mental_models = []
+
+    # When fact_type=["mental_model"], results come back in `results` field as MemoryFact objects
+    # We need to fetch additional fields (proof_count, source_memory_ids) from the database
+    if result.results:
+        mm_ids = [m.id for m in result.results]
+
+        # Fetch proof_count and source_memory_ids for these mental models
+        pool = await memory_engine._get_pool()
+        async with pool.acquire() as conn:
+            mm_rows = await conn.fetch(
+                f"""
+                SELECT id, proof_count, source_memory_ids
+                FROM {fq_table("memory_units")}
+                WHERE id = ANY($1::uuid[])
                 """,
-                bank_id,
+                mm_ids,
+            )
+            mm_data = {str(row["id"]): row for row in mm_rows}
+
+        for m in result.results:
+            # Get additional data from DB lookup
+            extra = mm_data.get(m.id, {})
+            proof_count = extra.get("proof_count", 1) if extra else 1
+            source_ids = extra.get("source_memory_ids", []) if extra else []
+            # Convert UUIDs to strings
+            source_memory_ids = [str(sid) for sid in (source_ids or [])]
+
+            # Determine staleness
+            is_stale = False
+            staleness_reason = None
+            if pending_consolidation > 0:
+                is_stale = True
+                staleness_reason = f"{pending_consolidation} memories pending consolidation"
+
+            mental_models.append(
+                {
+                    "id": str(m.id),
+                    "text": m.text,
+                    "proof_count": proof_count,
+                    "source_memory_ids": source_memory_ids,
+                    "tags": m.tags or [],
+                    "is_stale": is_stale,
+                    "staleness_reason": staleness_reason,
+                }
             )
 
-        return {
-            "count": len(rows),
-            "models": [
-                {
-                    "id": row["id"],
-                    "subtype": row["subtype"],
-                    "name": row["name"],
-                    "description": row["description"],
-                }
-                for row in rows
-            ],
-        }
+    # Return freshness info (more understandable than raw pending_consolidation count)
+    if pending_consolidation == 0:
+        freshness = "up_to_date"
+    elif pending_consolidation < 10:
+        freshness = "slightly_stale"
+    else:
+        freshness = "stale"
+
+    return {
+        "query": query,
+        "count": len(mental_models),
+        "mental_models": mental_models,
+        "freshness": freshness,
+    }
 
 
 async def tool_recall(
@@ -185,6 +246,9 @@ async def tool_recall(
     """
     Search memories using TEMPR retrieval.
 
+    This is the ground truth - raw facts and experiences.
+    Use when reflections/mental models don't exist, are stale, or need verification.
+
     Args:
         memory_engine: Memory engine instance
         bank_id: Bank identifier
@@ -202,13 +266,14 @@ async def tool_recall(
     result = await memory_engine.recall_async(
         bank_id=bank_id,
         query=query,
-        fact_type=["experience", "world"],  # Exclude opinions
+        fact_type=["experience", "world"],  # Exclude opinions and mental_models
         max_tokens=max_tokens,
         enable_trace=False,
         request_context=request_context,
         tags=tags,
         tags_match=tags_match,
         _connection_budget=connection_budget,
+        _quiet=True,  # Suppress logging for internal operations
     )
 
     memories = []
@@ -230,85 +295,6 @@ async def tool_recall(
     }
 
 
-async def tool_learn(
-    conn: "Connection",
-    bank_id: str,
-    input: MentalModelInput,
-    tags: list[str] | None = None,
-) -> dict[str, Any]:
-    """
-    Create a mental model placeholder with subtype='learned'.
-
-    The agent only specifies name and description - actual observations are generated
-    in the background via refresh, similar to pinned models.
-
-    Args:
-        conn: Database connection
-        bank_id: Bank identifier
-        input: Mental model input data (name, description, optional entity_id)
-        tags: Tags to apply to new mental models (from reflect context)
-
-    Returns:
-        Dict with created model info including model_id for background generation
-    """
-    model_id = generate_model_id(input.name)
-
-    # Parse entity_id if provided
-    entity_uuid = None
-    if input.entity_id:
-        try:
-            entity_uuid = uuid.UUID(input.entity_id)
-        except ValueError:
-            logger.warning(f"Invalid entity_id format: {input.entity_id}")
-
-    # Check if model exists
-    existing = await conn.fetchrow(
-        "SELECT id FROM mental_models WHERE id = $1 AND bank_id = $2",
-        model_id,
-        bank_id,
-    )
-
-    if existing:
-        # Update description only - observations will be regenerated
-        await conn.execute(
-            """
-            UPDATE mental_models SET
-                description = $3,
-                entity_id = $4
-            WHERE id = $1 AND bank_id = $2
-            """,
-            model_id,
-            bank_id,
-            input.description,
-            entity_uuid,
-        )
-        status = "updated"
-    else:
-        # Insert new model placeholder - observations will be generated in background
-        await conn.execute(
-            """
-            INSERT INTO mental_models (id, bank_id, subtype, name, description, observations, entity_id, tags, created_at)
-            VALUES ($1, $2, 'learned', $3, $4, '{}'::jsonb, $5, $6, NOW())
-            """,
-            model_id,
-            bank_id,
-            input.name,
-            input.description,
-            entity_uuid,
-            tags or [],
-        )
-        status = "created"
-
-    logger.info(f"[REFLECT] Mental model '{model_id}' {status} in bank {bank_id} - pending background generation")
-
-    return {
-        "status": status,
-        "model_id": model_id,
-        "name": input.name,
-        "pending_generation": True,
-    }
-
-
 async def tool_expand(
     conn: "Connection",
     bank_id: str,
@@ -327,6 +313,8 @@ async def tool_expand(
     Returns:
         Dict with results array, each containing memory, chunk, and optionally document data
     """
+    from ..memory_engine import fq_table
+
     if not memory_ids:
         return {"error": "memory_ids is required and must not be empty"}
 
@@ -344,9 +332,9 @@ async def tool_expand(
 
     # Batch fetch all memory units
     memories = await conn.fetch(
-        """
+        f"""
         SELECT id, text, chunk_id, document_id, fact_type, context
-        FROM memory_units
+        FROM {fq_table("memory_units")}
         WHERE id = ANY($1) AND bank_id = $2
         """,
         valid_uuids,
@@ -363,9 +351,9 @@ async def tool_expand(
     chunk_map: dict[str, Any] = {}
     if chunk_ids:
         chunks = await conn.fetch(
-            """
+            f"""
             SELECT chunk_id, chunk_text, chunk_index, document_id
-            FROM chunks
+            FROM {fq_table("chunks")}
             WHERE chunk_id = ANY($1)
             """,
             chunk_ids,
@@ -385,9 +373,9 @@ async def tool_expand(
     all_doc_ids = list(doc_ids_from_chunks | doc_ids_direct)
     if all_doc_ids:
         docs = await conn.fetch(
-            """
+            f"""
             SELECT id, original_text, metadata, retain_params
-            FROM documents
+            FROM {fq_table("documents")}
             WHERE id = ANY($1) AND bank_id = $2
             """,
             all_doc_ids,

hindsight-api/hindsight_api/engine/reflect/tools_schema.py

@@ -2,36 +2,62 @@
 Tool schema definitions for the reflect agent.
 
 These are OpenAI-format tool definitions used with native tool calling.
+The reflect agent uses a hierarchical retrieval strategy:
+1. search_reflections - User-curated summaries (highest quality, if applicable)
+2. search_mental_models - Consolidated knowledge with freshness awareness
+3. recall - Raw facts (world/experience) as ground truth fallback
 """
 
 # Tool definitions in OpenAI format
-TOOL_LIST_MENTAL_MODELS = {
+
+TOOL_SEARCH_REFLECTIONS = {
     "type": "function",
     "function": {
-        "name": "list_mental_models",
-        "description": "List all available mental models - your synthesized knowledge about entities, concepts, and events. Returns an array of models with id, name, and description.",
+        "name": "search_reflections",
+        "description": (
+            "Search user-curated reflections (summaries). These are high-quality, manually created "
+            "summaries about specific topics. Use FIRST when the question might be covered by an "
+            "existing reflection. Returns reflections with their content and last refresh time."
+        ),
         "parameters": {
             "type": "object",
-            "properties": {},
-            "required": [],
+            "properties": {
+                "query": {
+                    "type": "string",
+                    "description": "Search query to find relevant reflections",
+                },
+                "max_results": {
+                    "type": "integer",
+                    "description": "Maximum number of reflections to return (default 5)",
+                },
+            },
+            "required": ["query"],
         },
     },
 }
 
-TOOL_GET_MENTAL_MODEL = {
+TOOL_SEARCH_MENTAL_MODELS = {
     "type": "function",
     "function": {
-        "name": "get_mental_model",
-        "description": "Get full details of a specific mental model including all observations and memory references.",
+        "name": "search_mental_models",
+        "description": (
+            "Search consolidated mental models (auto-generated knowledge). These are automatically "
+            "synthesized from memories. Returns models with freshness info (updated_at, is_stale). "
+            "If a model is STALE, you should ALSO use recall() to verify with current facts."
+        ),
         "parameters": {
             "type": "object",
             "properties": {
-                "model_id": {
+                "query": {
                     "type": "string",
-                    "description": "ID of the mental model (from list_mental_models results)",
+                    "description": "Search query to find relevant mental models",
+                },
+                "max_tokens": {
+                    "type": "integer",
+                    "description": "Maximum tokens for results (default 5000). Use higher values for broader searches.",
                 },
             },
-            "required": ["model_id"],
+            "required": ["query"],
         },
     },
 }
@@ -40,7 +66,12 @@ TOOL_RECALL = {
     "type": "function",
     "function": {
         "name": "recall",
-        "description": "Search memories using semantic + temporal retrieval. Returns relevant memories from experience and world knowledge, each with an 'id' you can reference.",
+        "description": (
+            "Search raw memories (facts and experiences). This is the ground truth data. "
+            "Use when: (1) no reflections/mental models exist, (2) mental models are stale, "
+            "(3) you need specific details not in synthesized knowledge. "
+            "Returns individual memory facts with their timestamps."
+        ),
         "parameters": {
             "type": "object",
             "properties": {
@@ -58,28 +89,6 @@ TOOL_RECALL = {
     },
 }
 
-TOOL_LEARN = {
-    "type": "function",
-    "function": {
-        "name": "learn",
-        "description": "Create a new mental model to track an important recurring topic. Use when you discover a person, project, concept, or pattern that appears frequently and would benefit from synthesized knowledge. The model content will be generated automatically.",
-        "parameters": {
-            "type": "object",
-            "properties": {
-                "name": {
-                    "type": "string",
-                    "description": "Human-readable name (e.g., 'Project Alpha', 'John Smith', 'Product Strategy')",
-                },
-                "description": {
-                    "type": "string",
-                    "description": "What to track and synthesize (e.g., 'Track goals, milestones, blockers, and key decisions for Project Alpha')",
-                },
-            },
-            "required": ["name", "description"],
-        },
-    },
-}
-
 TOOL_EXPAND = {
     "type": "function",
     "function": {
@@ -121,7 +130,12 @@ TOOL_DONE_ANSWER = {
                     "items": {"type": "string"},
                     "description": "Array of memory IDs that support your answer (put IDs here, NOT in answer text)",
                 },
-                "model_ids": {
+                "reflection_ids": {
+                    "type": "array",
+                    "items": {"type": "string"},
+                    "description": "Array of reflection IDs that support your answer",
+                },
+                "mental_model_ids": {
                     "type": "array",
                     "items": {"type": "string"},
                     "description": "Array of mental model IDs that support your answer",
@@ -143,8 +157,6 @@ def _build_done_tool_with_directives(directive_rules: list[str]) -> dict:
     Args:
         directive_rules: List of directive rule strings
     """
-    from typing import Any, cast
-
     # Build rules list for description
     rules_list = "\n".join(f"  {i + 1}. {rule}" for i, rule in enumerate(directive_rules))
 
@@ -169,7 +181,12 @@ def _build_done_tool_with_directives(directive_rules: list[str]) -> dict:
                         "items": {"type": "string"},
                         "description": "Array of memory IDs that support your answer (put IDs here, NOT in answer text)",
                     },
-                    "model_ids": {
+                    "reflection_ids": {
+                        "type": "array",
+                        "items": {"type": "string"},
+                        "description": "Array of reflection IDs that support your answer",
+                    },
+                    "mental_model_ids": {
                         "type": "array",
                         "items": {"type": "string"},
                         "description": "Array of mental model IDs that support your answer",
@@ -185,29 +202,28 @@ def _build_done_tool_with_directives(directive_rules: list[str]) -> dict:
     }
 
 
-def get_reflect_tools(enable_learn: bool = True, directive_rules: list[str] | None = None) -> list[dict]:
+def get_reflect_tools(directive_rules: list[str] | None = None) -> list[dict]:
     """
     Get the list of tools for the reflect agent.
 
+    The tools support a hierarchical retrieval strategy:
+    1. search_reflections - User-curated summaries (try first)
+    2. search_mental_models - Consolidated knowledge with freshness
+    3. recall - Raw facts as ground truth
+
     Args:
-        enable_learn: Whether to include the learn tool
         directive_rules: Optional list of directive rule strings. If provided,
                         the done() tool will require directive compliance confirmation.
 
     Returns:
         List of tool definitions in OpenAI format
     """
-    tools = []
-
-    # Include mental model tools for lookup
-    tools.append(TOOL_LIST_MENTAL_MODELS)
-    tools.append(TOOL_GET_MENTAL_MODEL)
-    tools.append(TOOL_RECALL)
-
-    if enable_learn:
-        tools.append(TOOL_LEARN)
-
-    tools.append(TOOL_EXPAND)
+    tools = [
+        TOOL_SEARCH_REFLECTIONS,
+        TOOL_SEARCH_MENTAL_MODELS,
+        TOOL_RECALL,
+        TOOL_EXPAND,
+    ]
 
     # Use directive-aware done tool if directives are present
     if directive_rules:

hindsight-api/hindsight_api/engine/response_models.py

@@ -11,7 +11,7 @@ from typing import Any
 from pydantic import BaseModel, ConfigDict, Field
 
 # Valid fact types for recall operations (excludes 'observation' which is internal, and 'opinion' which is deprecated)
-VALID_RECALL_FACT_TYPES = frozenset(["world", "experience"])
+VALID_RECALL_FACT_TYPES = frozenset(["world", "experience", "mental_model"])
 
 
 class LLMToolCall(BaseModel):
@@ -166,6 +166,28 @@ class ChunkInfo(BaseModel):
     truncated: bool = Field(default=False, description="Whether the chunk was truncated due to token limits")
 
 
+class MentalModelResult(BaseModel):
+    """A mental model result from recall."""
+
+    id: str = Field(description="Unique mental model ID")
+    text: str = Field(description="The mental model text")
+    proof_count: int = Field(description="Number of facts supporting this mental model")
+    relevance: float = Field(default=0.0, description="Relevance score to the query")
+    tags: list[str] | None = Field(default=None, description="Tags for visibility scoping")
+    source_memory_ids: list[str] = Field(
+        default_factory=list, description="IDs of facts that contribute to this mental model"
+    )
+
+
+class ReflectionResult(BaseModel):
+    """A reflection result from recall."""
+
+    id: str = Field(description="Unique reflection ID")
+    name: str = Field(description="Human-readable name")
+    content: str = Field(description="The synthesized content")
+    relevance: float = Field(default=0.0, description="Relevance score to the query")
+
+
 class RecallResult(BaseModel):
     """
     Result from a recall operation.
@@ -229,6 +251,7 @@ class ReflectResult(BaseModel):
                     ],
                     "experience": [],
                     "opinion": [],
+                    "mental-models": [],
                 },
                 "new_opinions": ["Machine learning has great potential in healthcare"],
                 "structured_output": {"summary": "ML in healthcare", "confidence": 0.9},
@@ -239,7 +262,7 @@ class ReflectResult(BaseModel):
 
     text: str = Field(description="The formulated answer text")
     based_on: dict[str, list[MemoryFact]] = Field(
-        description="Facts used to formulate the answer, organized by type (world, experience, opinion)"
+        description="Facts used to formulate the answer, organized by type (world, experience, opinion, mental-models)"
     )
     new_opinions: list[str] = Field(default_factory=list, description="List of newly formed opinions during reflection")
     structured_output: dict[str, Any] | None = Field(
@@ -258,10 +281,6 @@ class ReflectResult(BaseModel):
         default_factory=list,
         description="Trace of LLM calls made during reflection. Only present when include.tool_calls is enabled.",
     )
-    mental_models: list[MentalModelRef] = Field(
-        default_factory=list,
-        description="Mental models accessed during reflection, including directives (subtype='directive').",
-    )
     directives_applied: list[DirectiveRef] = Field(
         default_factory=list,
         description="Directive mental models that were applied during this reflection.",

hindsight-api/hindsight_api/engine/retain/fact_extraction.py

@@ -114,11 +114,8 @@ class CausalRelation(BaseModel):
     """Causal relationship from this fact to a previous fact (stored format)."""
 
     target_fact_index: int = Field(description="Index of the related fact in the facts array (0-based).")
-    relation_type: Literal["caused_by", "enabled_by", "prevented_by"] = Field(
-        description="How this fact relates to the target: "
-        "'caused_by' = this fact was caused by the target, "
-        "'enabled_by' = this fact was enabled by the target, "
-        "'prevented_by' = this fact was prevented by the target"
+    relation_type: Literal["caused_by"] = Field(
+        description="How this fact relates to the target: 'caused_by' = this fact was caused by the target"
     )
     strength: float = Field(
         description="Strength of relationship (0.0 to 1.0)",
@@ -141,11 +138,8 @@ class FactCausalRelation(BaseModel):
         "MUST be less than this fact's position in the list. "
         "Example: if this is fact #5, target_index can only be 0, 1, 2, 3, or 4."
     )
-    relation_type: Literal["caused_by", "enabled_by", "prevented_by"] = Field(
-        description="How this fact relates to the target fact: "
-        "'caused_by' = this fact was caused by the target fact, "
-        "'enabled_by' = this fact was enabled by the target fact, "
-        "'prevented_by' = this fact was blocked/prevented by the target fact"
+    relation_type: Literal["caused_by"] = Field(
+        description="How this fact relates to the target fact: 'caused_by' = this fact was caused by the target fact"
     )
     strength: float = Field(
         description="Strength of relationship (0.0 to 1.0). 1.0 = strong, 0.5 = moderate",
@@ -441,7 +435,7 @@ def _chunk_conversation(turns: list[dict], max_chars: int) -> list[str]:
 # Concise extraction prompt (default) - selective, high-quality facts
 CONCISE_FACT_EXTRACTION_PROMPT = """Extract SIGNIFICANT facts from text. Be SELECTIVE - only extract facts worth remembering long-term.
 
-LANGUAGE RULE (CRITICAL): Output facts in the EXACT SAME language as the input text. If input is Japanese, output Japanese. If input is Chinese, output Chinese. NEVER translate to English. Preserve original language completely.
+LANGUAGE REQUIREMENT: Detect the language of the input text. All extracted facts, entity names, descriptions, and other output MUST be in the SAME language as the input. Do not translate to another language.
 
 {fact_types_instruction}
 
@@ -662,7 +656,7 @@ CAUSAL RELATIONSHIPS
 ══════════════════════════════════════════════════════════════════════════
 
 Link facts with causal_relations (max 2 per fact). target_index must be < this fact's index.
-Types: "caused_by", "enabled_by", "prevented_by"
+Type: "caused_by" (this fact was caused by the target fact)
 
 Example: "Lost job → couldn't pay rent → moved apartment"
 - Fact 0: Lost job, causal_relations: null
@@ -823,7 +817,8 @@ Text:
 
                 # Critical field: fact_type
                 # LLM uses "assistant" but we convert to "experience" for storage
-                fact_type = llm_fact.get("fact_type")
+                original_fact_type = llm_fact.get("fact_type")
+                fact_type = original_fact_type
 
                 # Convert "assistant" → "experience" for storage
                 if fact_type == "assistant":
@@ -840,7 +835,10 @@ Text:
                     else:
                         # Default to 'world' if we can't determine
                         fact_type = "world"
-                        logger.warning(f"Fact {i}: defaulting to fact_type='world'")
+                        logger.warning(
+                            f"Fact {i}: defaulting to fact_type='world' "
+                            f"(original fact_type={original_fact_type!r}, fact_kind={fact_kind!r})"
+                        )
 
                 # Get fact_kind for temporal handling (but don't store it)
                 fact_kind = llm_fact.get("fact_kind", "conversation")

hindsight-api/hindsight_api/engine/retain/fact_storage.py

@@ -41,7 +41,6 @@ async def insert_facts_batch(
     contexts = []
     fact_types = []
     confidence_scores = []
-    access_counts = []
     metadata_jsons = []
     chunk_ids = []
     document_ids = []
@@ -61,7 +60,6 @@ async def insert_facts_batch(
         fact_types.append(fact.fact_type)
         # confidence_score is only for opinion facts
         confidence_scores.append(1.0 if fact.fact_type == "opinion" else None)
-        access_counts.append(0)  # Initial access count
         metadata_jsons.append(json.dumps(fact.metadata))
         chunk_ids.append(fact.chunk_id)
         # Use per-fact document_id if available, otherwise fallback to batch-level document_id
@@ -76,16 +74,16 @@ async def insert_facts_batch(
         WITH input_data AS (
             SELECT * FROM unnest(
                 $2::text[], $3::vector[], $4::timestamptz[], $5::timestamptz[], $6::timestamptz[], $7::timestamptz[],
-                $8::text[], $9::text[], $10::float[], $11::int[], $12::jsonb[], $13::text[], $14::text[], $15::jsonb[]
+                $8::text[], $9::text[], $10::float[], $11::jsonb[], $12::text[], $13::text[], $14::jsonb[]
             ) AS t(text, embedding, event_date, occurred_start, occurred_end, mentioned_at,
-                   context, fact_type, confidence_score, access_count, metadata, chunk_id, document_id, tags_json)
+                   context, fact_type, confidence_score, metadata, chunk_id, document_id, tags_json)
         )
         INSERT INTO {fq_table("memory_units")} (bank_id, text, embedding, event_date, occurred_start, occurred_end, mentioned_at,
-                                 context, fact_type, confidence_score, access_count, metadata, chunk_id, document_id, tags)
+                                 context, fact_type, confidence_score, metadata, chunk_id, document_id, tags)
         SELECT
             $1,
             text, embedding, event_date, occurred_start, occurred_end, mentioned_at,
-            context, fact_type, confidence_score, access_count, metadata, chunk_id, document_id,
+            context, fact_type, confidence_score, metadata, chunk_id, document_id,
             COALESCE(
                 (SELECT array_agg(elem) FROM jsonb_array_elements_text(tags_json) AS elem),
                 '{{}}'::varchar[]
@@ -103,7 +101,6 @@ async def insert_facts_batch(
         contexts,
         fact_types,
         confidence_scores,
-        access_counts,
         metadata_jsons,
         chunk_ids,
         document_ids,

hindsight-api/hindsight_api/engine/retain/link_utils.py

@@ -754,17 +754,14 @@ async def create_causal_links_batch(
         causal_relations_per_fact: List of causal relations for each fact.
             Each element is a list of dicts with:
             - target_fact_index: Index into unit_ids for the target fact
-            - relation_type: "causes", "caused_by", "enables", or "prevents"
+            - relation_type: "caused_by"
             - strength: Float in [0.0, 1.0] representing relationship strength
 
     Returns:
         Number of causal links created
 
-    Causal link types:
-    - "causes": This fact directly causes the target fact (forward causation)
-    - "caused_by": This fact was caused by the target fact (backward causation)
-    - "enables": This fact enables/allows the target fact (enablement)
-    - "prevents": This fact prevents/blocks the target fact (prevention)
+    Causal link type:
+    - "caused_by": This fact was caused by the target fact
     """
     if not unit_ids or not causal_relations_per_fact:
         return 0
@@ -787,8 +784,8 @@ async def create_causal_links_batch(
                 relation_type = relation["relation_type"]
                 strength = relation.get("strength", 1.0)
 
-                # Validate relation_type - must match database constraint
-                valid_types = {"causes", "caused_by", "enables", "prevents"}
+                # Validate relation_type - only "caused_by" is supported (DB constraint)
+                valid_types = {"caused_by"}
                 if relation_type not in valid_types:
                     logger.error(
                         f"Invalid relation_type '{relation_type}' (type: {type(relation_type).__name__}) "

hindsight-api/hindsight_api/engine/retain/types.py

@@ -86,10 +86,10 @@ class CausalRelation:
     """
     Causal relationship between facts.
 
-    Represents how one fact causes, enables, or prevents another.
+    Represents how one fact was caused by another.
     """
 
-    relation_type: str  # "causes", "enables", "prevents", "caused_by"
+    relation_type: str  # "caused_by"
     target_fact_index: int  # Index of the target fact in the batch
     strength: float = 1.0  # Strength of the causal relationship
 

hindsight-api/hindsight_api/engine/search/graph_retrieval.py

@@ -162,7 +162,7 @@ class BFSGraphRetriever(GraphRetriever):
         entry_points = await conn.fetch(
             f"""
             SELECT id, text, context, event_date, occurred_start, occurred_end,
-                   mentioned_at, access_count, embedding, fact_type, document_id, chunk_id, tags,
+                   mentioned_at, embedding, fact_type, document_id, chunk_id, tags,
                    1 - (embedding <=> $1::vector) AS similarity
             FROM {fq_table("memory_units")}
             WHERE bank_id = $2
@@ -216,7 +216,7 @@ class BFSGraphRetriever(GraphRetriever):
                 neighbors = await conn.fetch(
                     f"""
                     SELECT mu.id, mu.text, mu.context, mu.occurred_start, mu.occurred_end,
-                           mu.mentioned_at, mu.access_count, mu.embedding, mu.fact_type,
+                           mu.mentioned_at, mu.embedding, mu.fact_type,
                            mu.document_id, mu.chunk_id, mu.tags,
                            ml.weight, ml.link_type, ml.from_unit_id
                     FROM {fq_table("memory_links")} ml

hindsight-api/hindsight_api/engine/search/link_expansion_retrieval.py

@@ -45,7 +45,7 @@ async def _find_semantic_seeds(
     rows = await conn.fetch(
         f"""
         SELECT id, text, context, event_date, occurred_start, occurred_end,
-               mentioned_at, access_count, embedding, fact_type, document_id, chunk_id, tags,
+               mentioned_at, embedding, fact_type, document_id, chunk_id, tags,
                1 - (embedding <=> $1::vector) AS similarity
         FROM {fq_table("memory_units")}
         WHERE bank_id = $2
@@ -168,7 +168,7 @@ class LinkExpansionRetriever(GraphRetriever):
                 f"""
                 SELECT
                     mu.id, mu.text, mu.context, mu.event_date, mu.occurred_start,
-                    mu.occurred_end, mu.mentioned_at, mu.access_count, mu.embedding,
+                    mu.occurred_end, mu.mentioned_at, mu.embedding,
                     mu.fact_type, mu.document_id, mu.chunk_id, mu.tags,
                     COUNT(*)::float AS score
                 FROM {fq_table("unit_entities")} seed_ue
@@ -193,7 +193,7 @@ class LinkExpansionRetriever(GraphRetriever):
                 f"""
                 SELECT DISTINCT ON (mu.id)
                     mu.id, mu.text, mu.context, mu.event_date, mu.occurred_start,
-                    mu.occurred_end, mu.mentioned_at, mu.access_count, mu.embedding,
+                    mu.occurred_end, mu.mentioned_at, mu.embedding,
                     mu.fact_type, mu.document_id, mu.chunk_id, mu.tags,
                     ml.weight + 1.0 AS score
                 FROM {fq_table("memory_links")} ml

hindsight-api/hindsight_api/engine/search/mpfp_retrieval.py

@@ -449,7 +449,7 @@ async def fetch_memory_units_by_ids(
         rows = await conn.fetch(
             f"""
             SELECT id, text, context, event_date, occurred_start, occurred_end,
-                   mentioned_at, access_count, embedding, fact_type, document_id, chunk_id, tags
+                   mentioned_at, embedding, fact_type, document_id, chunk_id, tags
             FROM {fq_table("memory_units")}
             WHERE id = ANY($1::uuid[])
               AND fact_type = $2

hindsight-api/hindsight_api/engine/search/retrieval.py

@@ -116,7 +116,7 @@ async def retrieve_semantic(
 
     results = await conn.fetch(
         f"""
-        SELECT id, text, context, event_date, occurred_start, occurred_end, mentioned_at, access_count, embedding, fact_type, document_id, chunk_id, tags,
+        SELECT id, text, context, event_date, occurred_start, occurred_end, mentioned_at, embedding, fact_type, document_id, chunk_id, tags,
                1 - (embedding <=> $1::vector) AS similarity
         FROM {fq_table("memory_units")}
         WHERE bank_id = $2
@@ -180,7 +180,7 @@ async def retrieve_bm25(
 
     results = await conn.fetch(
         f"""
-        SELECT id, text, context, event_date, occurred_start, occurred_end, mentioned_at, access_count, embedding, fact_type, document_id, chunk_id, tags,
+        SELECT id, text, context, event_date, occurred_start, occurred_end, mentioned_at, embedding, fact_type, document_id, chunk_id, tags,
                ts_rank_cd(search_vector, to_tsquery('english', $1)) AS bm25_score
         FROM {fq_table("memory_units")}
         WHERE bank_id = $2
@@ -237,7 +237,7 @@ async def retrieve_semantic_bm25_combined(
         results = await conn.fetch(
             f"""
             WITH semantic_ranked AS (
-                SELECT id, text, context, event_date, occurred_start, occurred_end, mentioned_at, access_count, embedding, fact_type, document_id, chunk_id, tags,
+                SELECT id, text, context, event_date, occurred_start, occurred_end, mentioned_at, embedding, fact_type, document_id, chunk_id, tags,
                        1 - (embedding <=> $1::vector) AS similarity,
                        NULL::float AS bm25_score,
                        'semantic' AS source,
@@ -249,7 +249,7 @@ async def retrieve_semantic_bm25_combined(
                   AND (1 - (embedding <=> $1::vector)) >= 0.3
                   {tags_clause}
             )
-            SELECT id, text, context, event_date, occurred_start, occurred_end, mentioned_at, access_count, embedding, fact_type, document_id, chunk_id, tags,
+            SELECT id, text, context, event_date, occurred_start, occurred_end, mentioned_at, embedding, fact_type, document_id, chunk_id, tags,
                    similarity, bm25_score, source
             FROM semantic_ranked
             WHERE rn <= $4
@@ -281,7 +281,7 @@ async def retrieve_semantic_bm25_combined(
     results = await conn.fetch(
         f"""
         WITH semantic_ranked AS (
-            SELECT id, text, context, event_date, occurred_start, occurred_end, mentioned_at, access_count, embedding, fact_type, document_id, chunk_id, tags,
+            SELECT id, text, context, event_date, occurred_start, occurred_end, mentioned_at, embedding, fact_type, document_id, chunk_id, tags,
                    1 - (embedding <=> $1::vector) AS similarity,
                    NULL::float AS bm25_score,
                    'semantic' AS source,
@@ -294,7 +294,7 @@ async def retrieve_semantic_bm25_combined(
               {tags_clause}
         ),
         bm25_ranked AS (
-            SELECT id, text, context, event_date, occurred_start, occurred_end, mentioned_at, access_count, embedding, fact_type, document_id, chunk_id, tags,
+            SELECT id, text, context, event_date, occurred_start, occurred_end, mentioned_at, embedding, fact_type, document_id, chunk_id, tags,
                    NULL::float AS similarity,
                    ts_rank_cd(search_vector, to_tsquery('english', $5)) AS bm25_score,
                    'bm25' AS source,
@@ -306,12 +306,12 @@ async def retrieve_semantic_bm25_combined(
               {tags_clause}
         ),
         semantic AS (
-            SELECT id, text, context, event_date, occurred_start, occurred_end, mentioned_at, access_count, embedding, fact_type, document_id, chunk_id, tags,
+            SELECT id, text, context, event_date, occurred_start, occurred_end, mentioned_at, embedding, fact_type, document_id, chunk_id, tags,
                    similarity, bm25_score, source
             FROM semantic_ranked WHERE rn <= $4
         ),
         bm25 AS (
-            SELECT id, text, context, event_date, occurred_start, occurred_end, mentioned_at, access_count, embedding, fact_type, document_id, chunk_id, tags,
+            SELECT id, text, context, event_date, occurred_start, occurred_end, mentioned_at, embedding, fact_type, document_id, chunk_id, tags,
                    similarity, bm25_score, source
             FROM bm25_ranked WHERE rn <= $4
         )
@@ -386,7 +386,7 @@ async def retrieve_temporal_combined(
     entry_points = await conn.fetch(
         f"""
         WITH ranked_entries AS (
-            SELECT id, text, context, event_date, occurred_start, occurred_end, mentioned_at, access_count, embedding, fact_type, document_id, chunk_id, tags,
+            SELECT id, text, context, event_date, occurred_start, occurred_end, mentioned_at, embedding, fact_type, document_id, chunk_id, tags,
                    1 - (embedding <=> $1::vector) AS similarity,
                    ROW_NUMBER() OVER (PARTITION BY fact_type ORDER BY COALESCE(occurred_start, mentioned_at, occurred_end) DESC, embedding <=> $1::vector) AS rn
             FROM {fq_table("memory_units")}
@@ -406,7 +406,7 @@ async def retrieve_temporal_combined(
               AND (1 - (embedding <=> $1::vector)) >= $6
               {tags_clause}
         )
-        SELECT id, text, context, event_date, occurred_start, occurred_end, mentioned_at, access_count, embedding, fact_type, document_id, chunk_id, tags, similarity
+        SELECT id, text, context, event_date, occurred_start, occurred_end, mentioned_at, embedding, fact_type, document_id, chunk_id, tags, similarity
         FROM ranked_entries
         WHERE rn <= 10
         """,
@@ -486,7 +486,7 @@ async def retrieve_temporal_combined(
 
             neighbors = await conn.fetch(
                 f"""
-                SELECT mu.id, mu.text, mu.context, mu.event_date, mu.occurred_start, mu.occurred_end, mu.mentioned_at, mu.access_count, mu.embedding, mu.fact_type, mu.document_id, mu.chunk_id, mu.tags,
+                SELECT mu.id, mu.text, mu.context, mu.event_date, mu.occurred_start, mu.occurred_end, mu.mentioned_at, mu.embedding, mu.fact_type, mu.document_id, mu.chunk_id, mu.tags,
                        ml.weight, ml.link_type, ml.from_unit_id,
                        1 - (mu.embedding <=> $1::vector) AS similarity
                 FROM {fq_table("memory_links")} ml
@@ -610,7 +610,7 @@ async def retrieve_temporal(
 
     entry_points = await conn.fetch(
         f"""
-        SELECT id, text, context, event_date, occurred_start, occurred_end, mentioned_at, access_count, embedding, fact_type, document_id, chunk_id, tags,
+        SELECT id, text, context, event_date, occurred_start, occurred_end, mentioned_at, embedding, fact_type, document_id, chunk_id, tags,
                1 - (embedding <=> $1::vector) AS similarity
         FROM {fq_table("memory_units")}
         WHERE bank_id = $2
@@ -691,7 +691,7 @@ async def retrieve_temporal(
         # Batch fetch all neighbors for this batch of nodes
         neighbors = await conn.fetch(
             f"""
-            SELECT mu.id, mu.text, mu.context, mu.event_date, mu.occurred_start, mu.occurred_end, mu.mentioned_at, mu.access_count, mu.embedding, mu.fact_type, mu.document_id, mu.chunk_id,
+            SELECT mu.id, mu.text, mu.context, mu.event_date, mu.occurred_start, mu.occurred_end, mu.mentioned_at, mu.embedding, mu.fact_type, mu.document_id, mu.chunk_id,
                    ml.weight, ml.link_type, ml.from_unit_id,
                    1 - (mu.embedding <=> $1::vector) AS similarity
             FROM {fq_table("memory_links")} ml
@@ -1023,7 +1023,7 @@ async def _get_temporal_entry_points(
     rows = await conn.fetch(
         f"""
         SELECT id, text, context, event_date, occurred_start, occurred_end, mentioned_at,
-               access_count, embedding, fact_type, document_id, chunk_id,
+               embedding, fact_type, document_id, chunk_id,
                1 - (embedding <=> $1::vector) AS similarity
         FROM {fq_table("memory_units")}
         WHERE bank_id = $2

hindsight-api/hindsight_api/engine/search/scoring.py

@@ -65,31 +65,6 @@ def calculate_recency_weight(days_since: float, half_life_days: float = 365.0) -
     return 1.0 / (1.0 + math.log1p(normalized_age))
 
 
-def calculate_frequency_weight(access_count: int, max_boost: float = 2.0) -> float:
-    """
-    Calculate frequency weight based on access count.
-
-    Frequently accessed memories are weighted higher.
-    Uses logarithmic scaling to avoid over-weighting.
-
-    Args:
-        access_count: Number of times the memory was accessed
-        max_boost: Maximum multiplier for frequently accessed memories
-
-    Returns:
-        Weight between 1.0 and max_boost
-    """
-    import math
-
-    if access_count <= 0:
-        return 1.0
-
-    # Logarithmic scaling: log(access_count + 1) / log(10)
-    # This gives: 0 accesses = 1.0, 9 accesses ~= 1.5, 99 accesses ~= 2.0
-    normalized = math.log(access_count + 1) / math.log(10)
-    return 1.0 + min(normalized, max_boost - 1.0)
-
-
 def calculate_temporal_anchor(occurred_start: datetime, occurred_end: datetime) -> datetime:
     """
     Calculate a single temporal anchor point from a temporal range.

hindsight-api/hindsight_api/engine/search/trace.py

@@ -85,7 +85,6 @@ class NodeVisit(BaseModel):
     text: str = Field(description="Memory unit text content")
     context: str = Field(description="Memory unit context")
     event_date: datetime | None = Field(default=None, description="When the memory occurred")
-    access_count: int = Field(description="Number of times accessed before this search")
 
     # How this node was reached
     is_entry_point: bool = Field(description="Whether this is an entry point")

hindsight-api/hindsight_api/engine/search/tracer.py

@@ -136,7 +136,6 @@ class SearchTracer:
         text: str,
         context: str,
         event_date: datetime | None,
-        access_count: int,
         is_entry_point: bool,
         parent_node_id: str | None,
         link_type: Literal["temporal", "semantic", "entity"] | None,
@@ -155,7 +154,6 @@ class SearchTracer:
             text: Memory unit text
             context: Memory unit context
             event_date: When the memory occurred
-            access_count: Access count before this search
             is_entry_point: Whether this is an entry point
             parent_node_id: Node that led here (None for entry points)
             link_type: Type of link from parent
@@ -194,7 +192,6 @@ class SearchTracer:
             text=text,
             context=context,
             event_date=event_date,
-            access_count=access_count,
             is_entry_point=is_entry_point,
             parent_node_id=parent_node_id,
             link_type=link_type,
@@ -333,8 +330,8 @@ class SearchTracer:
                 RetrievalResult(
                     rank=rank,
                     node_id=doc_id,
-                    text=data.get("text", ""),
-                    context=data.get("context", ""),
+                    text=data.get("text") or "",
+                    context=data.get("context") or "",
                     event_date=data.get("event_date"),
                     fact_type=data.get("fact_type") or fact_type,
                     score=score,

hindsight-api/hindsight_api/engine/search/types.py

@@ -46,7 +46,6 @@ class RetrievalResult:
     mentioned_at: datetime | None = None
     document_id: str | None = None
     chunk_id: str | None = None
-    access_count: int = 0
     embedding: list[float] | None = None
     tags: list[str] | None = None  # Visibility scope tags
 
@@ -71,7 +70,6 @@ class RetrievalResult:
             mentioned_at=row.get("mentioned_at"),
             document_id=row.get("document_id"),
             chunk_id=row.get("chunk_id"),
-            access_count=row.get("access_count", 0),
             embedding=row.get("embedding"),
             tags=row.get("tags"),
             similarity=row.get("similarity"),
@@ -156,7 +154,6 @@ class ScoredResult:
             "mentioned_at": self.retrieval.mentioned_at,
             "document_id": self.retrieval.document_id,
             "chunk_id": self.retrieval.chunk_id,
-            "access_count": self.retrieval.access_count,
             "embedding": self.retrieval.embedding,
             "tags": self.retrieval.tags,
             "semantic_similarity": self.retrieval.similarity,

hindsight-api/hindsight_api/engine/task_backend.py

@@ -1,31 +1,40 @@
 """
-Abstract task backend for running async tasks.
+Task backend for distributed task processing.
 
-This provides an abstraction that can be adapted to different execution models:
-- AsyncIO queue (default implementation)
-- Pub/Sub architectures (future)
-- Message brokers (future)
+This provides an abstraction for task storage and execution:
+- BrokerTaskBackend: Uses PostgreSQL as broker (production)
+- SyncTaskBackend: Executes tasks immediately (testing/embedded)
 """
 
-import asyncio
+import json
 import logging
 from abc import ABC, abstractmethod
 from collections.abc import Awaitable, Callable
-from typing import Any
+from typing import TYPE_CHECKING, Any
+
+if TYPE_CHECKING:
+    import asyncpg
 
 logger = logging.getLogger(__name__)
 
 
+def fq_table(table: str, schema: str | None = None) -> str:
+    """Get fully-qualified table name with optional schema prefix."""
+    if schema:
+        return f'"{schema}".{table}'
+    return table
+
+
 class TaskBackend(ABC):
     """
     Abstract base class for task execution backends.
 
     Implementations must:
     1. Store/publish task events (as serializable dicts)
-    2. Execute tasks through a provided executor callback
+    2. Execute tasks through a provided executor callback (optional)
 
     The backend treats tasks as pure dictionaries that can be serialized
-    and sent over the network. The executor (typically MemoryEngine.execute_task)
+    and stored in the database. The executor (typically MemoryEngine.execute_task)
     receives the dict and routes it to the appropriate handler.
     """
 
@@ -46,7 +55,7 @@ class TaskBackend(ABC):
     @abstractmethod
     async def initialize(self):
         """
-        Initialize the backend (e.g., start workers, connect to broker).
+        Initialize the backend (e.g., connect to database).
         """
         pass
 
@@ -63,7 +72,7 @@ class TaskBackend(ABC):
     @abstractmethod
     async def shutdown(self):
         """
-        Shutdown the backend gracefully (e.g., stop workers, close connections).
+        Shutdown the backend gracefully.
         """
         pass
 
@@ -93,9 +102,8 @@ class SyncTaskBackend(TaskBackend):
     """
     Synchronous task backend that executes tasks immediately.
 
-    This is useful for embedded/CLI usage where we don't want background
-    workers that prevent clean exit. Tasks are executed inline rather than
-    being queued.
+    This is useful for tests and embedded/CLI usage where we don't want
+    background workers. Tasks are executed inline rather than being queued.
     """
 
     async def initialize(self):
@@ -121,221 +129,123 @@ class SyncTaskBackend(TaskBackend):
         logger.debug("SyncTaskBackend shutdown")
 
 
-class NoopTaskBackend(TaskBackend):
+class BrokerTaskBackend(TaskBackend):
     """
-    No-op task backend that discards all tasks.
+    Task backend using PostgreSQL as broker.
 
-    This is useful for tests where background task execution is not needed
-    and would only slow down the test suite.
+    submit_task() stores task_payload in async_operations table.
+    Actual polling and execution is handled separately by WorkerPoller.
+
+    This backend is used by the API to store tasks. Workers poll
+    the database separately to claim and execute tasks.
     """
 
-    async def initialize(self):
-        """No-op."""
-        self._initialized = True
-        logger.debug("NoopTaskBackend initialized")
-
-    async def submit_task(self, task_dict: dict[str, Any]):
-        """Discard the task (do nothing)."""
-        pass
-
-    async def shutdown(self):
-        """No-op."""
-        self._initialized = False
-        logger.debug("NoopTaskBackend shutdown")
-
-
-class AsyncIOQueueBackend(TaskBackend):
-    """
-    Task backend implementation using asyncio queues.
-
-    This is the default implementation that uses in-process asyncio queues
-    and a periodic consumer worker.
-    """
-
-    def __init__(self, batch_size: int = 10, batch_interval: float = 1.0):
+    def __init__(
+        self,
+        pool_getter: Callable[[], "asyncpg.Pool"],
+        schema: str | None = None,
+    ):
         """
-        Initialize AsyncIO queue backend.
+        Initialize the broker task backend.
 
         Args:
-            batch_size: Maximum number of tasks to process in one batch
-            batch_interval: Maximum time (seconds) to wait before processing batch
+            pool_getter: Callable that returns the asyncpg connection pool
+            schema: Database schema for multi-tenant support (optional)
         """
         super().__init__()
-        self._queue: asyncio.Queue | None = None
-        self._worker_task: asyncio.Task | None = None
-        self._shutdown_event: asyncio.Event | None = None
-        self._batch_size = batch_size
-        self._batch_interval = batch_interval
-        self._in_flight_count = 0
-        self._in_flight_lock = asyncio.Lock()
+        self._pool_getter = pool_getter
+        self._schema = schema
 
     async def initialize(self):
-        """Initialize the queue and start the worker."""
-        if self._initialized:
-            return
-
-        self._queue = asyncio.Queue()
-        self._shutdown_event = asyncio.Event()
-        self._worker_task = asyncio.create_task(self._worker())
+        """Initialize the backend."""
         self._initialized = True
-        logger.info("AsyncIOQueueBackend initialized")
+        logger.info("BrokerTaskBackend initialized")
 
     async def submit_task(self, task_dict: dict[str, Any]):
         """
-        Submit a task by putting it in the queue.
+        Store task payload in async_operations table.
+
+        The task_dict should contain an 'operation_id' if updating an existing
+        operation record, otherwise a new operation will be created.
 
         Args:
-            task_dict: Task dictionary to execute
+            task_dict: Task dictionary to store (must be JSON serializable)
         """
         if not self._initialized:
             await self.initialize()
 
-        await self._queue.put(task_dict)
+        pool = self._pool_getter()
+        operation_id = task_dict.get("operation_id")
+        task_type = task_dict.get("type", "unknown")
+        bank_id = task_dict.get("bank_id")
+        payload_json = json.dumps(task_dict)
+
+        table = fq_table("async_operations", self._schema)
+
+        if operation_id:
+            # Update existing operation with task payload
+            await pool.execute(
+                f"""
+                UPDATE {table}
+                SET task_payload = $1::jsonb, updated_at = now()
+                WHERE operation_id = $2
+                """,
+                payload_json,
+                operation_id,
+            )
+            logger.debug(f"Updated task payload for operation {operation_id}")
+        else:
+            # Insert new operation (for tasks without pre-created records)
+            # e.g., access_count_update tasks
+            import uuid
+
+            new_id = uuid.uuid4()
+            await pool.execute(
+                f"""
+                INSERT INTO {table} (operation_id, bank_id, operation_type, status, task_payload)
+                VALUES ($1, $2, $3, 'pending', $4::jsonb)
+                """,
+                new_id,
+                bank_id,
+                task_type,
+                payload_json,
+            )
+            logger.debug(f"Created new operation {new_id} for task type {task_type}")
+
+    async def shutdown(self):
+        """Shutdown the backend."""
+        self._initialized = False
+        logger.info("BrokerTaskBackend shutdown")
 
     async def wait_for_pending_tasks(self, timeout: float = 120.0):
         """
-        Wait for all pending tasks in the queue and in-flight tasks to complete.
+        Wait for pending tasks to be processed.
 
-        This is useful in tests to ensure background tasks complete before assertions.
+        In the broker model, this polls the database to check if tasks
+        for this process have been completed. This is useful in tests
+        when worker_enabled=True (API processes its own tasks).
 
         Args:
-            timeout: Maximum time to wait in seconds (default 120s for long-running tasks)
+            timeout: Maximum time to wait in seconds
         """
-        if not self._initialized or self._queue is None:
-            return
+        import asyncio
+
+        pool = self._pool_getter()
+        table = fq_table("async_operations", self._schema)
 
-        # Wait for queue to be empty AND no in-flight tasks
         start_time = asyncio.get_event_loop().time()
         while asyncio.get_event_loop().time() - start_time < timeout:
-            async with self._in_flight_lock:
-                in_flight = self._in_flight_count
+            # Check if there are any pending tasks with payloads
+            count = await pool.fetchval(
+                f"""
+                SELECT COUNT(*) FROM {table}
+                WHERE status = 'pending' AND task_payload IS NOT NULL
+                """
+            )
 
-            if self._queue.empty() and in_flight == 0:
-                # Queue is empty and no tasks in flight, we're done
+            if count == 0:
                 return
 
-            # Wait a bit before checking again
             await asyncio.sleep(0.5)
 
-    async def shutdown(self):
-        """Shutdown the worker and drain the queue."""
-        if not self._initialized:
-            return
-
-        logger.info("Shutting down AsyncIOQueueBackend...")
-
-        # Signal shutdown
-        self._shutdown_event.set()
-
-        # Cancel worker
-        if self._worker_task is not None:
-            self._worker_task.cancel()
-            try:
-                await self._worker_task
-            except asyncio.CancelledError:
-                pass  # Worker cancelled successfully
-
-        self._initialized = False
-        logger.info("AsyncIOQueueBackend shutdown complete")
-
-    async def _execute_task_with_tracking(self, task_dict: dict[str, Any]):
-        """Execute a task and track its in-flight status."""
-        async with self._in_flight_lock:
-            self._in_flight_count += 1
-        try:
-            await self._execute_task(task_dict)
-        finally:
-            async with self._in_flight_lock:
-                self._in_flight_count -= 1
-
-    async def _execute_task_no_tracking(self, task_dict: dict[str, Any]):
-        """Execute a task without in-flight tracking (tracking done at batch level)."""
-        await self._execute_task(task_dict)
-
-    def _get_queue_stats(self) -> tuple[int, dict[str, int]]:
-        """Get current queue size and bank_id distribution."""
-        queue_size = self._queue.qsize() if self._queue else 0
-        bank_distribution: dict[str, int] = {}
-
-        if queue_size > 0 and self._queue:
-            # Peek at queue items without removing them
-            # Note: This is a snapshot and may not be perfectly accurate due to concurrency
-            try:
-                # Access internal deque for logging purposes only
-                items = list(self._queue._queue)  # type: ignore[attr-defined]
-                for item in items:
-                    bank_id = item.get("bank_id", "unknown")
-                    bank_distribution[bank_id] = bank_distribution.get(bank_id, 0) + 1
-            except Exception:
-                pass  # Queue access failed, return empty distribution
-
-        return queue_size, bank_distribution
-
-    async def _worker(self):
-        """
-        Background worker that processes tasks in batches.
-
-        Collects tasks for up to batch_interval seconds or batch_size items,
-        then processes them.
-        """
-        while not self._shutdown_event.is_set():
-            try:
-                # Collect tasks for batching
-                tasks = []
-                deadline = asyncio.get_event_loop().time() + self._batch_interval
-
-                while len(tasks) < self._batch_size and asyncio.get_event_loop().time() < deadline:
-                    try:
-                        remaining_time = max(0.1, deadline - asyncio.get_event_loop().time())
-                        task_dict = await asyncio.wait_for(self._queue.get(), timeout=remaining_time)
-                        # Track task as in-flight immediately when picked up from queue
-                        # This prevents wait_for_pending_tasks from returning too early
-                        async with self._in_flight_lock:
-                            self._in_flight_count += 1
-                        tasks.append(task_dict)
-                    except TimeoutError:
-                        break
-
-                # Process batch
-                if tasks:
-                    # Log batch start with queue stats
-                    queue_size, bank_distribution = self._get_queue_stats()
-
-                    # Summarize batch by task type and bank
-                    batch_summary: dict[str, dict[str, int]] = {}
-                    for task_dict in tasks:
-                        task_type = task_dict.get("type", "unknown")
-                        bank_id = task_dict.get("bank_id", "unknown")
-                        if task_type not in batch_summary:
-                            batch_summary[task_type] = {}
-                        batch_summary[task_type][bank_id] = batch_summary[task_type].get(bank_id, 0) + 1
-
-                    # Build log message
-                    batch_parts = []
-                    for task_type, banks in sorted(batch_summary.items()):
-                        bank_str = ", ".join(f"{b}:{c}" for b, c in sorted(banks.items()))
-                        batch_parts.append(f"{task_type}[{bank_str}]")
-                    batch_str = ", ".join(batch_parts)
-
-                    if queue_size > 0:
-                        pending_str = ", ".join(f"{k}:{v}" for k, v in sorted(bank_distribution.items()))
-                        logger.info(
-                            f"Processing {len(tasks)} tasks: {batch_str} (pending={queue_size} [{pending_str}])"
-                        )
-                    else:
-                        logger.info(f"Processing {len(tasks)} tasks: {batch_str}")
-
-                    # Execute tasks concurrently (in_flight already tracked when picked up)
-                    await asyncio.gather(
-                        *[self._execute_task_no_tracking(task_dict) for task_dict in tasks], return_exceptions=True
-                    )
-
-                    # Decrement in_flight count after all tasks complete
-                    async with self._in_flight_lock:
-                        self._in_flight_count -= len(tasks)
-
-            except asyncio.CancelledError:
-                break
-            except Exception as e:
-                logger.error(f"Worker error: {e}")
-                await asyncio.sleep(1)  # Backoff on error
+        logger.warning(f"Timeout waiting for pending tasks after {timeout}s")

hindsight-api/hindsight_api/engine/utils.py

@@ -124,31 +124,6 @@ def calculate_recency_weight(days_since: float, half_life_days: float = 365.0) -
     return 1.0 / (1.0 + math.log1p(normalized_age))
 
 
-def calculate_frequency_weight(access_count: int, max_boost: float = 2.0) -> float:
-    """
-    Calculate frequency weight based on access count.
-
-    Frequently accessed memories are weighted higher.
-    Uses logarithmic scaling to avoid over-weighting.
-
-    Args:
-        access_count: Number of times the memory was accessed
-        max_boost: Maximum multiplier for frequently accessed memories
-
-    Returns:
-        Weight between 1.0 and max_boost
-    """
-    import math
-
-    if access_count <= 0:
-        return 1.0
-
-    # Logarithmic scaling: log(access_count + 1) / log(10)
-    # This gives: 0 accesses = 1.0, 9 accesses ~= 1.5, 99 accesses ~= 2.0
-    normalized = math.log(access_count + 1) / math.log(10)
-    return 1.0 + min(normalized, max_boost - 1.0)
-
-
 def calculate_temporal_anchor(occurred_start: datetime, occurred_end: datetime) -> datetime:
     """
     Calculate a single temporal anchor point from a temporal range.

hindsight-api/hindsight_api/extensions/__init__.py

@@ -27,8 +27,6 @@ from hindsight_api.extensions.operation_validator import (
     RecallResult,
     ReflectContext,
     ReflectResultContext,
-    RefreshMentalModelContext,
-    RefreshMentalModelResult,
     RetainContext,
     RetainResult,
     ValidationResult,
@@ -56,8 +54,6 @@ __all__ = [
     "RecallResult",
     "ReflectContext",
     "ReflectResultContext",
-    "RefreshMentalModelContext",
-    "RefreshMentalModelResult",
     "RetainContext",
     "RetainResult",
     "ValidationResult",

hindsight-api/hindsight_api/extensions/operation_validator.py

@@ -97,18 +97,6 @@ class ReflectContext:
     context: str | None = None
 
 
-@dataclass
-class RefreshMentalModelContext:
-    """Context for a refresh mental model operation validation (pre-operation).
-
-    Contains ALL user-provided parameters for the refresh mental model operation.
-    """
-
-    bank_id: str
-    model_id: str
-    request_context: "RequestContext"
-
-
 # =============================================================================
 # Post-operation Contexts (includes results)
 # =============================================================================
@@ -176,27 +164,6 @@ class ReflectResultContext:
     error: str | None = None
 
 
-@dataclass
-class RefreshMentalModelResult:
-    """Result context for post-refresh-mental-model hook.
-
-    Contains the operation parameters and the result including token usage.
-    """
-
-    bank_id: str
-    model_id: str
-    request_context: "RequestContext"
-    # Result
-    model_name: str | None = None
-    observations_count: int = 0
-    input_tokens: int = 0
-    output_tokens: int = 0
-    total_tokens: int = 0
-    duration_ms: int = 0
-    success: bool = True
-    error: str | None = None
-
-
 class OperationValidatorExtension(Extension, ABC):
     """
     Validates and hooks into retain/recall/reflect operations.
@@ -298,25 +265,6 @@ class OperationValidatorExtension(Extension, ABC):
         """
         ...
 
-    @abstractmethod
-    async def validate_refresh_mental_model(self, ctx: RefreshMentalModelContext) -> ValidationResult:
-        """
-        Validate a refresh mental model operation before execution.
-
-        Called before the refresh mental model operation is processed.
-        Return ValidationResult.reject() to prevent the operation from executing.
-
-        Args:
-            ctx: Context containing all user-provided parameters:
-                - bank_id: Bank identifier
-                - model_id: Mental model ID to refresh
-                - request_context: Request context with auth info
-
-        Returns:
-            ValidationResult indicating whether the operation is allowed.
-        """
-        ...
-
     # =========================================================================
     # Post-operation hooks (optional - override to implement)
     # =========================================================================
@@ -377,28 +325,3 @@ class OperationValidatorExtension(Extension, ABC):
                 - error: Error message (if failed)
         """
         pass
-
-    async def on_refresh_mental_model_complete(self, result: RefreshMentalModelResult) -> None:
-        """
-        Called after a refresh mental model operation completes (success or failure).
-
-        Override this method to implement post-operation logic such as:
-        - Token usage tracking and billing
-        - Audit logging
-        - Metrics collection
-
-        Args:
-            result: Result context containing:
-                - bank_id: Bank identifier
-                - model_id: Mental model ID
-                - request_context: Request context with auth info
-                - model_name: Name of the mental model (if success)
-                - observations_count: Number of observations generated
-                - input_tokens: Number of input tokens used
-                - output_tokens: Number of output tokens used
-                - total_tokens: Total tokens used (input + output)
-                - duration_ms: Total operation duration in milliseconds
-                - success: Whether the operation succeeded
-                - error: Error message (if failed)
-        """
-        pass

hindsight-api/hindsight_api/main.py

@@ -212,6 +212,9 @@ def main():
             retain_extract_causal_links=config.retain_extract_causal_links,
             retain_extraction_mode=config.retain_extraction_mode,
             retain_observations_async=config.retain_observations_async,
+            enable_mental_models=config.enable_mental_models,
+            consolidation_similarity_threshold=config.consolidation_similarity_threshold,
+            consolidation_batch_size=config.consolidation_batch_size,
             skip_llm_verification=config.skip_llm_verification,
             lazy_reranker=config.lazy_reranker,
             run_migrations_on_startup=config.run_migrations_on_startup,
@@ -219,9 +222,12 @@ def main():
             db_pool_max_size=config.db_pool_max_size,
             db_command_timeout=config.db_command_timeout,
             db_acquire_timeout=config.db_acquire_timeout,
-            task_backend=config.task_backend,
-            task_backend_memory_batch_size=config.task_backend_memory_batch_size,
-            task_backend_memory_batch_interval=config.task_backend_memory_batch_interval,
+            worker_enabled=config.worker_enabled,
+            worker_id=config.worker_id,
+            worker_poll_interval_ms=config.worker_poll_interval_ms,
+            worker_max_retries=config.worker_max_retries,
+            worker_batch_size=config.worker_batch_size,
+            worker_http_port=config.worker_http_port,
             reflect_max_iterations=config.reflect_max_iterations,
             mental_model_refresh_concurrency=config.mental_model_refresh_concurrency,
         )

hindsight-api/hindsight_api/models.py

@@ -95,7 +95,6 @@ class MemoryUnit(Base):
     mentioned_at: Mapped[datetime | None] = mapped_column(TIMESTAMP(timezone=True))  # When fact was mentioned
     fact_type: Mapped[str] = mapped_column(Text, nullable=False, server_default="world")
     confidence_score: Mapped[float | None] = mapped_column(Float)
-    access_count: Mapped[int] = mapped_column(Integer, server_default="0")
     unit_metadata: Mapped[dict] = mapped_column(
         "metadata", JSONB, server_default=sql_text("'{}'::jsonb")
     )  # User-defined metadata (str->str)
@@ -131,7 +130,6 @@ class MemoryUnit(Base):
         Index("idx_memory_units_document_id", "document_id"),
         Index("idx_memory_units_event_date", "event_date", postgresql_ops={"event_date": "DESC"}),
         Index("idx_memory_units_bank_date", "bank_id", "event_date", postgresql_ops={"event_date": "DESC"}),
-        Index("idx_memory_units_access_count", "access_count", postgresql_ops={"access_count": "DESC"}),
         Index("idx_memory_units_fact_type", "fact_type"),
         Index("idx_memory_units_bank_fact_type", "bank_id", "fact_type"),
         Index(

hindsight-api/hindsight_api/worker/__init__.py

@@ -0,0 +1,11 @@
+"""
+Worker package for distributed task processing.
+
+This package provides:
+- WorkerPoller: Polls PostgreSQL for pending tasks and executes them
+- main: CLI entry point for hindsight-worker
+"""
+
+from .poller import WorkerPoller
+
+__all__ = ["WorkerPoller"]

hindsight-api/hindsight_api/worker/main.py

@@ -0,0 +1,285 @@
+"""
+Command-line interface for Hindsight Worker.
+
+Run the worker with:
+    hindsight-worker
+
+Stop with Ctrl+C (graceful shutdown).
+"""
+
+import argparse
+import asyncio
+import atexit
+import logging
+import os
+import signal
+import socket
+import sys
+import warnings
+
+from ..config import get_config
+from ..engine.task_backend import SyncTaskBackend
+from .poller import WorkerPoller
+
+# Filter deprecation warnings from third-party libraries
+warnings.filterwarnings("ignore", message="websockets.legacy is deprecated")
+warnings.filterwarnings("ignore", message="websockets.server.WebSocketServerProtocol is deprecated")
+
+# Disable tokenizers parallelism to avoid warnings
+os.environ["TOKENIZERS_PARALLELISM"] = "false"
+
+logger = logging.getLogger(__name__)
+
+
+def create_worker_app(poller: WorkerPoller, memory):
+    """Create a minimal FastAPI app for worker metrics and health."""
+    from fastapi import FastAPI
+    from fastapi.responses import JSONResponse, Response
+    from prometheus_client import CONTENT_TYPE_LATEST, generate_latest
+
+    from ..metrics import create_metrics_collector, get_metrics_collector, initialize_metrics
+
+    app = FastAPI(
+        title="Hindsight Worker",
+        description="Worker process for distributed task execution",
+    )
+
+    # Initialize OpenTelemetry metrics
+    try:
+        prometheus_reader = initialize_metrics(service_name="hindsight-worker", service_version="1.0.0")
+        create_metrics_collector()
+        app.state.prometheus_reader = prometheus_reader
+        logger.info("Metrics initialized - available at /metrics endpoint")
+    except Exception as e:
+        logger.warning(f"Failed to initialize metrics: {e}. Metrics will be disabled.")
+        app.state.prometheus_reader = None
+
+    # Set up DB pool metrics if available
+    metrics_collector = get_metrics_collector()
+    if memory._pool is not None and hasattr(metrics_collector, "set_db_pool"):
+        metrics_collector.set_db_pool(memory._pool)
+        logger.info("DB pool metrics configured")
+
+    @app.get(
+        "/health",
+        summary="Health check endpoint",
+        description="Returns worker health status including database connectivity",
+        tags=["Monitoring"],
+    )
+    async def health_endpoint():
+        """Health check endpoint."""
+        health = await memory.health_check()
+        health["worker_id"] = poller.worker_id
+        health["is_shutdown"] = poller.is_shutdown
+        status_code = 200 if health.get("status") == "healthy" else 503
+        return JSONResponse(content=health, status_code=status_code)
+
+    @app.get(
+        "/metrics",
+        summary="Prometheus metrics endpoint",
+        description="Exports metrics in Prometheus format for scraping",
+        tags=["Monitoring"],
+    )
+    async def metrics_endpoint():
+        """Return Prometheus metrics."""
+        metrics_data = generate_latest()
+        return Response(content=metrics_data, media_type=CONTENT_TYPE_LATEST)
+
+    @app.get(
+        "/",
+        summary="Worker info",
+        description="Basic worker information",
+        tags=["Info"],
+    )
+    async def root():
+        """Return basic worker info."""
+        return {
+            "service": "hindsight-worker",
+            "worker_id": poller.worker_id,
+            "is_shutdown": poller.is_shutdown,
+        }
+
+    return app
+
+
+def main():
+    """Main entry point for the hindsight-worker CLI."""
+    # Load configuration from environment
+    config = get_config()
+
+    parser = argparse.ArgumentParser(
+        prog="hindsight-worker",
+        description="Hindsight Worker - distributed task processor",
+    )
+
+    # Worker options
+    parser.add_argument(
+        "--worker-id",
+        default=config.worker_id or socket.gethostname(),
+        help="Worker identifier (default: hostname, env: HINDSIGHT_API_WORKER_ID)",
+    )
+    parser.add_argument(
+        "--poll-interval",
+        type=int,
+        default=config.worker_poll_interval_ms,
+        help=f"Poll interval in milliseconds (default: {config.worker_poll_interval_ms}, env: HINDSIGHT_API_WORKER_POLL_INTERVAL_MS)",
+    )
+    parser.add_argument(
+        "--batch-size",
+        type=int,
+        default=config.worker_batch_size,
+        help=f"Tasks to claim per poll (default: {config.worker_batch_size}, env: HINDSIGHT_API_WORKER_BATCH_SIZE)",
+    )
+    parser.add_argument(
+        "--max-retries",
+        type=int,
+        default=config.worker_max_retries,
+        help=f"Max retries before marking failed (default: {config.worker_max_retries}, env: HINDSIGHT_API_WORKER_MAX_RETRIES)",
+    )
+
+    # HTTP server options
+    parser.add_argument(
+        "--http-port",
+        type=int,
+        default=config.worker_http_port,
+        help=f"HTTP port for metrics/health endpoints (default: {config.worker_http_port}, env: HINDSIGHT_API_WORKER_HTTP_PORT)",
+    )
+    parser.add_argument(
+        "--http-host",
+        default="0.0.0.0",
+        help="HTTP host to bind (default: 0.0.0.0)",
+    )
+
+    # Logging options
+    parser.add_argument(
+        "--log-level",
+        default=config.log_level,
+        choices=["critical", "error", "warning", "info", "debug", "trace"],
+        help=f"Log level (default: {config.log_level}, env: HINDSIGHT_API_LOG_LEVEL)",
+    )
+
+    args = parser.parse_args()
+
+    # Configure logging
+    config.configure_logging()
+
+    # Import MemoryEngine here to avoid circular imports
+    from .. import MemoryEngine
+
+    print(f"Starting Hindsight Worker: {args.worker_id}")
+    print(f"  Poll interval: {args.poll_interval}ms")
+    print(f"  Batch size: {args.batch_size}")
+    print(f"  Max retries: {args.max_retries}")
+    print(f"  HTTP server: {args.http_host}:{args.http_port}")
+    print()
+
+    # Global references for cleanup
+    memory = None
+    poller = None
+
+    async def run():
+        nonlocal memory, poller
+        import uvicorn
+
+        # Initialize MemoryEngine
+        # Workers use SyncTaskBackend because they execute tasks directly,
+        # they don't need to store tasks (they poll from DB)
+        memory = MemoryEngine(
+            run_migrations=False,  # Workers don't run migrations
+            task_backend=SyncTaskBackend(),
+        )
+
+        await memory.initialize()
+
+        print(f"Database connected: {config.database_url}")
+
+        # Create and start the poller
+        poller = WorkerPoller(
+            pool=memory._pool,
+            worker_id=args.worker_id,
+            executor=memory.execute_task,
+            poll_interval_ms=args.poll_interval,
+            batch_size=args.batch_size,
+            max_retries=args.max_retries,
+        )
+
+        # Create the HTTP app for metrics/health
+        app = create_worker_app(poller, memory)
+
+        # Setup signal handlers for graceful shutdown
+        shutdown_requested = asyncio.Event()
+
+        def signal_handler(signum, frame):
+            print(f"\nReceived signal {signum}, initiating graceful shutdown...")
+            shutdown_requested.set()
+
+        signal.signal(signal.SIGINT, signal_handler)
+        signal.signal(signal.SIGTERM, signal_handler)
+
+        # Create uvicorn config and server
+        uvicorn_config = uvicorn.Config(
+            app,
+            host=args.http_host,
+            port=args.http_port,
+            log_level="info",  # Reduce uvicorn noise
+            access_log=False,
+        )
+        server = uvicorn.Server(uvicorn_config)
+
+        # Run the poller and HTTP server concurrently
+        poller_task = asyncio.create_task(poller.run())
+        http_task = asyncio.create_task(server.serve())
+
+        print(f"Worker started. Metrics available at http://{args.http_host}:{args.http_port}/metrics")
+
+        # Wait for shutdown signal
+        await shutdown_requested.wait()
+
+        # Graceful shutdown
+        print("Shutting down HTTP server...")
+        server.should_exit = True
+
+        print("Waiting for poller to finish...")
+        await poller.shutdown_graceful(timeout=30.0)
+        poller_task.cancel()
+        try:
+            await poller_task
+        except asyncio.CancelledError:
+            pass
+
+        # Wait for HTTP server to finish
+        try:
+            await asyncio.wait_for(http_task, timeout=5.0)
+        except asyncio.TimeoutError:
+            http_task.cancel()
+            try:
+                await http_task
+            except asyncio.CancelledError:
+                pass
+
+        # Close memory engine
+        await memory.close()
+        print("Worker shutdown complete")
+
+    def cleanup():
+        """Synchronous cleanup for atexit."""
+        if memory is not None and memory._pg0 is not None:
+            try:
+                loop = asyncio.new_event_loop()
+                loop.run_until_complete(memory._pg0.stop())
+                loop.close()
+                print("\npg0 stopped.")
+            except Exception as e:
+                print(f"\nError stopping pg0: {e}")
+
+    atexit.register(cleanup)
+
+    try:
+        asyncio.run(run())
+    except KeyboardInterrupt:
+        print("\nWorker interrupted")
+        sys.exit(0)
+
+
+if __name__ == "__main__":
+    main()

hindsight-api/hindsight_api/worker/poller.py

@@ -0,0 +1,404 @@
+"""
+Worker poller for distributed task execution.
+
+Polls PostgreSQL for pending tasks and executes them using
+FOR UPDATE SKIP LOCKED for safe concurrent claiming.
+"""
+
+import asyncio
+import json
+import logging
+import time
+import traceback
+from collections.abc import Awaitable, Callable
+from typing import TYPE_CHECKING, Any
+
+if TYPE_CHECKING:
+    import asyncpg
+
+logger = logging.getLogger(__name__)
+
+# Progress logging interval in seconds
+PROGRESS_LOG_INTERVAL = 30
+
+
+def fq_table(table: str, schema: str | None = None) -> str:
+    """Get fully-qualified table name with optional schema prefix."""
+    if schema:
+        return f'"{schema}".{table}'
+    return table
+
+
+class WorkerPoller:
+    """
+    Polls PostgreSQL for pending tasks and executes them.
+
+    Uses FOR UPDATE SKIP LOCKED for safe distributed claiming,
+    allowing multiple workers to process tasks without conflicts.
+    """
+
+    def __init__(
+        self,
+        pool: "asyncpg.Pool",
+        worker_id: str,
+        executor: Callable[[dict[str, Any]], Awaitable[None]],
+        poll_interval_ms: int = 500,
+        batch_size: int = 10,
+        max_retries: int = 3,
+        schema: str | None = None,
+    ):
+        """
+        Initialize the worker poller.
+
+        Args:
+            pool: asyncpg connection pool
+            worker_id: Unique identifier for this worker
+            executor: Async function to execute tasks (typically MemoryEngine.execute_task)
+            poll_interval_ms: Interval between polls when no tasks found (milliseconds)
+            batch_size: Maximum number of tasks to claim per poll cycle
+            max_retries: Maximum retry attempts before marking task as failed
+            schema: Database schema for multi-tenant support (optional)
+        """
+        self._pool = pool
+        self._worker_id = worker_id
+        self._executor = executor
+        self._poll_interval_ms = poll_interval_ms
+        self._batch_size = batch_size
+        self._max_retries = max_retries
+        self._schema = schema
+        self._shutdown = asyncio.Event()
+        self._current_tasks: set[asyncio.Task] = set()
+        self._in_flight_count = 0
+        self._in_flight_lock = asyncio.Lock()
+        self._last_progress_log = 0.0
+        self._tasks_completed_since_log = 0
+        self._active_banks: set[str] = set()
+
+    async def claim_batch(self) -> list[tuple[str, dict[str, Any]]]:
+        """
+        Claim up to batch_size pending tasks atomically.
+
+        Uses FOR UPDATE SKIP LOCKED to ensure no conflicts with other workers.
+
+        For consolidation tasks specifically, skips pending tasks if there's already
+        a processing consolidation for the same bank (to avoid duplicate work).
+
+        Returns:
+            List of tuples (operation_id, task_dict)
+        """
+        table = fq_table("async_operations", self._schema)
+
+        async with self._pool.acquire() as conn:
+            async with conn.transaction():
+                # Select and lock pending tasks
+                # For consolidation: skip if same bank already has one processing
+                rows = await conn.fetch(
+                    f"""
+                    SELECT operation_id, task_payload
+                    FROM {table} AS pending
+                    WHERE status = 'pending' AND task_payload IS NOT NULL
+                    AND (
+                        -- Non-consolidation tasks: always claimable
+                        operation_type != 'consolidation'
+                        OR
+                        -- Consolidation: only if no other consolidation processing for same bank
+                        NOT EXISTS (
+                            SELECT 1 FROM {table} AS processing
+                            WHERE processing.bank_id = pending.bank_id
+                            AND processing.operation_type = 'consolidation'
+                            AND processing.status = 'processing'
+                        )
+                    )
+                    ORDER BY created_at
+                    LIMIT $1
+                    FOR UPDATE SKIP LOCKED
+                    """,
+                    self._batch_size,
+                )
+
+                if not rows:
+                    return []
+
+                # Claim the tasks by updating status and worker_id
+                operation_ids = [row["operation_id"] for row in rows]
+                await conn.execute(
+                    f"""
+                    UPDATE {table}
+                    SET status = 'processing', worker_id = $1, claimed_at = now(), updated_at = now()
+                    WHERE operation_id = ANY($2)
+                    """,
+                    self._worker_id,
+                    operation_ids,
+                )
+
+                # Parse and return task payloads
+                return [(str(row["operation_id"]), json.loads(row["task_payload"])) for row in rows]
+
+    async def _mark_completed(self, operation_id: str):
+        """Mark a task as completed."""
+        table = fq_table("async_operations", self._schema)
+        await self._pool.execute(
+            f"""
+            UPDATE {table}
+            SET status = 'completed', completed_at = now(), updated_at = now()
+            WHERE operation_id = $1
+            """,
+            operation_id,
+        )
+
+    async def _mark_failed(self, operation_id: str, error_message: str):
+        """Mark a task as failed with error message."""
+        table = fq_table("async_operations", self._schema)
+        # Truncate error message if too long (max 5000 chars in schema)
+        error_message = error_message[:5000] if len(error_message) > 5000 else error_message
+        await self._pool.execute(
+            f"""
+            UPDATE {table}
+            SET status = 'failed', error_message = $2, completed_at = now(), updated_at = now()
+            WHERE operation_id = $1
+            """,
+            operation_id,
+            error_message,
+        )
+
+    async def _retry_or_fail(self, operation_id: str, error_message: str):
+        """Increment retry count or mark as failed if max retries exceeded."""
+        table = fq_table("async_operations", self._schema)
+
+        # Get current retry count
+        row = await self._pool.fetchrow(
+            f"SELECT retry_count FROM {table} WHERE operation_id = $1",
+            operation_id,
+        )
+
+        if row is None:
+            logger.warning(f"Operation {operation_id} not found, cannot retry")
+            return
+
+        retry_count = row["retry_count"]
+
+        if retry_count >= self._max_retries:
+            # Max retries exceeded, mark as failed
+            await self._mark_failed(
+                operation_id, f"Max retries ({self._max_retries}) exceeded. Last error: {error_message}"
+            )
+            logger.error(f"Task {operation_id} failed after {retry_count} retries")
+        else:
+            # Increment retry and reset to pending
+            await self._pool.execute(
+                f"""
+                UPDATE {table}
+                SET status = 'pending', worker_id = NULL, claimed_at = NULL,
+                    retry_count = retry_count + 1, updated_at = now()
+                WHERE operation_id = $1
+                """,
+                operation_id,
+            )
+            logger.warning(f"Task {operation_id} failed, will retry (attempt {retry_count + 1}/{self._max_retries})")
+
+    async def execute_task(self, operation_id: str, task_dict: dict[str, Any]):
+        """Execute a single task and update its status."""
+        task_type = task_dict.get("type", "unknown")
+        bank_id = task_dict.get("bank_id", "unknown")
+
+        try:
+            logger.debug(f"Executing task {operation_id} (type={task_type}, bank={bank_id})")
+            await self._executor(task_dict)
+            await self._mark_completed(operation_id)
+            logger.debug(f"Task {operation_id} completed successfully")
+        except Exception as e:
+            error_msg = f"{type(e).__name__}: {e}\n{traceback.format_exc()}"
+            logger.error(f"Task {operation_id} failed: {e}")
+            await self._retry_or_fail(operation_id, error_msg)
+
+    async def recover_own_tasks(self) -> int:
+        """
+        Recover tasks that were assigned to this worker but not completed.
+
+        This handles the case where a worker crashes while processing tasks.
+        On startup, we reset any tasks stuck in 'processing' for this worker_id
+        back to 'pending' so they can be picked up again.
+
+        Returns:
+            Number of tasks recovered
+        """
+        table = fq_table("async_operations", self._schema)
+
+        result = await self._pool.execute(
+            f"""
+            UPDATE {table}
+            SET status = 'pending', worker_id = NULL, claimed_at = NULL, updated_at = now()
+            WHERE status = 'processing' AND worker_id = $1
+            """,
+            self._worker_id,
+        )
+
+        # Parse "UPDATE N" to get count
+        count = int(result.split()[-1]) if result else 0
+        if count > 0:
+            logger.info(f"Worker {self._worker_id} recovered {count} stale tasks from previous run")
+        return count
+
+    async def run(self):
+        """
+        Main polling loop.
+
+        Continuously polls for pending tasks, claims them, and executes them
+        until shutdown is signaled.
+        """
+        # Recover any tasks from a previous crash before starting
+        await self.recover_own_tasks()
+
+        logger.info(f"Worker {self._worker_id} starting polling loop")
+
+        while not self._shutdown.is_set():
+            try:
+                # Claim a batch of tasks
+                tasks = await self.claim_batch()
+
+                if tasks:
+                    # Log batch info
+                    task_types = {}
+                    for _, task_dict in tasks:
+                        t = task_dict.get("type", "unknown")
+                        task_types[t] = task_types.get(t, 0) + 1
+                    types_str = ", ".join(f"{k}:{v}" for k, v in task_types.items())
+                    logger.info(f"Worker {self._worker_id} claimed {len(tasks)} tasks: {types_str}")
+
+                    # Track in-flight tasks
+                    async with self._in_flight_lock:
+                        self._in_flight_count += len(tasks)
+
+                    # Execute tasks concurrently
+                    try:
+                        await asyncio.gather(
+                            *[self.execute_task(op_id, task_dict) for op_id, task_dict in tasks],
+                            return_exceptions=True,
+                        )
+                    finally:
+                        async with self._in_flight_lock:
+                            self._in_flight_count -= len(tasks)
+                else:
+                    # No tasks found, wait before polling again
+                    try:
+                        await asyncio.wait_for(
+                            self._shutdown.wait(),
+                            timeout=self._poll_interval_ms / 1000,
+                        )
+                    except asyncio.TimeoutError:
+                        pass  # Normal timeout, continue polling
+
+                # Log progress stats periodically
+                await self._log_progress_if_due()
+
+            except asyncio.CancelledError:
+                logger.info(f"Worker {self._worker_id} polling loop cancelled")
+                break
+            except Exception as e:
+                logger.error(f"Worker {self._worker_id} error in polling loop: {e}")
+                traceback.print_exc()
+                # Backoff on error
+                await asyncio.sleep(1)
+
+        logger.info(f"Worker {self._worker_id} polling loop stopped")
+
+    async def shutdown_graceful(self, timeout: float = 30.0):
+        """
+        Signal shutdown and wait for current tasks to complete.
+
+        Args:
+            timeout: Maximum time to wait for in-flight tasks (seconds)
+        """
+        logger.info(f"Worker {self._worker_id} initiating graceful shutdown")
+        self._shutdown.set()
+
+        # Wait for in-flight tasks to complete
+        start_time = asyncio.get_event_loop().time()
+        while asyncio.get_event_loop().time() - start_time < timeout:
+            async with self._in_flight_lock:
+                in_flight = self._in_flight_count
+
+            if in_flight == 0:
+                logger.info(f"Worker {self._worker_id} graceful shutdown complete")
+                return
+
+            logger.info(f"Worker {self._worker_id} waiting for {in_flight} in-flight tasks")
+            await asyncio.sleep(0.5)
+
+        logger.warning(f"Worker {self._worker_id} shutdown timeout after {timeout}s")
+
+    async def _log_progress_if_due(self):
+        """Log progress stats every PROGRESS_LOG_INTERVAL seconds."""
+        now = time.time()
+        if now - self._last_progress_log < PROGRESS_LOG_INTERVAL:
+            return
+
+        self._last_progress_log = now
+
+        try:
+            table = fq_table("async_operations", self._schema)
+            async with self._pool.acquire() as conn:
+                # Get global stats by status
+                stats = await conn.fetch(
+                    f"""
+                    SELECT status, COUNT(*) as count
+                    FROM {table}
+                    WHERE created_at > now() - interval '24 hours'
+                    GROUP BY status
+                    """
+                )
+
+                # Get currently processing tasks grouped by type and bank
+                processing = await conn.fetch(
+                    f"""
+                    SELECT operation_type, bank_id, COUNT(*) as count
+                    FROM {table}
+                    WHERE status = 'processing'
+                    GROUP BY operation_type, bank_id
+                    """
+                )
+
+            # Build stats dict
+            status_counts = {row["status"]: row["count"] for row in stats}
+            pending = status_counts.get("pending", 0)
+            processing_count = status_counts.get("processing", 0)
+            completed = status_counts.get("completed", 0)
+            failed = status_counts.get("failed", 0)
+
+            # Build processing breakdown
+            processing_info = []
+            banks_working = set()
+            for row in processing:
+                op_type = row["operation_type"]
+                bank_id = row["bank_id"]
+                count = row["count"]
+                banks_working.add(bank_id)
+                processing_info.append(f"{op_type}:{bank_id}({count})")
+
+            # Format log
+            async with self._in_flight_lock:
+                in_flight = self._in_flight_count
+
+            processing_str = ", ".join(processing_info[:10]) if processing_info else "none"
+            if len(processing_info) > 10:
+                processing_str += f" +{len(processing_info) - 10} more"
+
+            logger.info(
+                f"[WORKER_STATS] worker={self._worker_id} in_flight={in_flight} | "
+                f"global: pending={pending} processing={processing_count} "
+                f"completed_24h={completed} failed_24h={failed} | "
+                f"active: {processing_str}"
+            )
+
+        except Exception as e:
+            logger.debug(f"Failed to log progress stats: {e}")
+
+    @property
+    def worker_id(self) -> str:
+        """Get the worker ID."""
+        return self._worker_id
+
+    @property
+    def is_shutdown(self) -> bool:
+        """Check if shutdown has been signaled."""
+        return self._shutdown.is_set()

hindsight-api/pyproject.toml

@@ -63,6 +63,7 @@ test = [
 
 [project.scripts]
 hindsight-api = "hindsight_api.main:main"
+hindsight-worker = "hindsight_api.worker.main:main"
 hindsight-local-mcp = "hindsight_api.mcp_local:main"
 hindsight-admin = "hindsight_api.admin.cli:main"
 

hindsight-api/tests/conftest.py

@@ -12,6 +12,7 @@ from hindsight_api import MemoryEngine, LLMConfig, LocalSTEmbeddings, RequestCon
 
 from hindsight_api.engine.cross_encoder import LocalSTCrossEncoder
 from hindsight_api.engine.query_analyzer import DateparserQueryAnalyzer
+from hindsight_api.engine.task_backend import SyncTaskBackend
 from hindsight_api.pg0 import EmbeddedPostgres
 
 # Default pg0 instance configuration for tests
@@ -115,16 +116,65 @@ def llm_config():
 
 
 @pytest.fixture(scope="session")
-def embeddings():
+def embeddings(tmp_path_factory, worker_id):
+    """
+    Session-scoped embeddings fixture with filelock to prevent race conditions.
 
-    return LocalSTEmbeddings()
+    When pytest-xdist runs multiple workers in parallel, they all try to load
+    models from the HuggingFace cache simultaneously, which can cause race
+    conditions and meta tensor errors. We use a filelock to serialize model
+    initialization across workers.
+    """
+    # Get shared temp dir for coordination between xdist workers
+    if worker_id == "master":
+        root_tmp_dir = tmp_path_factory.getbasetemp()
+    else:
+        root_tmp_dir = tmp_path_factory.getbasetemp().parent
 
+    lock_file = root_tmp_dir / "embeddings_init.lock"
+
+    emb = LocalSTEmbeddings()
+
+    # Serialize model initialization across workers
+    with filelock.FileLock(str(lock_file)):
+        loop = asyncio.new_event_loop()
+        try:
+            loop.run_until_complete(emb.initialize())
+        finally:
+            loop.close()
+
+    return emb
 
 
 @pytest.fixture(scope="session")
-def cross_encoder():
+def cross_encoder(tmp_path_factory, worker_id):
+    """
+    Session-scoped cross-encoder fixture with filelock to prevent race conditions.
 
-    return LocalSTCrossEncoder()
+    When pytest-xdist runs multiple workers in parallel, they all try to load
+    models from the HuggingFace cache simultaneously, which can cause race
+    conditions and meta tensor errors. We use a filelock to serialize model
+    initialization across workers.
+    """
+    # Get shared temp dir for coordination between xdist workers
+    if worker_id == "master":
+        root_tmp_dir = tmp_path_factory.getbasetemp()
+    else:
+        root_tmp_dir = tmp_path_factory.getbasetemp().parent
+
+    lock_file = root_tmp_dir / "cross_encoder_init.lock"
+
+    ce = LocalSTCrossEncoder()
+
+    # Serialize model initialization across workers
+    with filelock.FileLock(str(lock_file)):
+        loop = asyncio.new_event_loop()
+        try:
+            loop.run_until_complete(ce.initialize())
+        finally:
+            loop.close()
+
+    return ce
 
 @pytest.fixture(scope="session")
 def query_analyzer():
@@ -147,6 +197,7 @@ async def memory(pg0_db_url, embeddings, cross_encoder, query_analyzer):
     Uses pg0_db_url (a postgresql:// URL) directly, so MemoryEngine won't try to
     manage pg0 lifecycle - that's handled by the session-scoped pg0_db_url fixture.
     Migrations are disabled here since they're run once at session scope in pg0_db_url.
+    Uses SyncTaskBackend so async tasks execute immediately (no worker needed).
     """
     mem = MemoryEngine(
         db_url=pg0_db_url,  # Direct postgresql:// URL, not pg0://
@@ -160,6 +211,7 @@ async def memory(pg0_db_url, embeddings, cross_encoder, query_analyzer):
         pool_min_size=1,
         pool_max_size=5,
         run_migrations=False,  # Migrations already run at session scope
+        task_backend=SyncTaskBackend(),  # Execute tasks immediately in tests
     )
     await mem.initialize()
     yield mem

hindsight-api/tests/test_custom_embedding_dimension.py

@@ -17,6 +17,7 @@ from hindsight_api import MemoryEngine, RequestContext
 from hindsight_api.engine.embeddings import LocalSTEmbeddings, OpenAIEmbeddings, CohereEmbeddings
 from hindsight_api.engine.cross_encoder import LocalSTCrossEncoder, CohereCrossEncoder
 from hindsight_api.engine.query_analyzer import DateparserQueryAnalyzer
+from hindsight_api.engine.task_backend import SyncTaskBackend
 from hindsight_api.extensions import TenantExtension, TenantContext
 from hindsight_api.migrations import run_migrations, ensure_embedding_dimension
 
@@ -323,6 +324,7 @@ class TestOpenAIEmbeddings:
             pool_max_size=3,
             run_migrations=False,
             tenant_extension=SchemaTenantExtension(schema_name),
+            task_backend=SyncTaskBackend(),
         )
 
         try:
@@ -392,6 +394,7 @@ class TestOpenAIEmbeddings:
             pool_max_size=3,
             run_migrations=False,
             tenant_extension=SchemaTenantExtension(schema_name),
+            task_backend=SyncTaskBackend(),
         )
 
         try:
@@ -559,6 +562,7 @@ class TestCohereIntegration:
             pool_max_size=3,
             run_migrations=False,
             tenant_extension=SchemaTenantExtension(schema_name),
+            task_backend=SyncTaskBackend(),
         )
 
         try:

hindsight-api/tests/test_emergent_filtering.py

@@ -1,516 +0,0 @@
-"""Tests for emergent entity filtering."""
-
-import pytest
-from unittest.mock import AsyncMock, MagicMock
-
-from hindsight_api.engine.mental_models.emergent import (
-    build_mission_filter_prompt,
-    evaluate_emergent_models,
-    filter_candidates_by_mission,
-    MissionFilterResponse,
-    MissionFilterCandidate,
-)
-from hindsight_api.engine.mental_models.models import EmergentCandidate
-
-
-class TestBuildMissionFilterPrompt:
-    """Test prompt building for mission filtering."""
-
-    def test_prompt_contains_mission(self):
-        """Test that prompt includes the mission."""
-        candidates = [
-            EmergentCandidate(
-                name="Alice",
-                detection_method="named_entity_extraction",
-                mention_count=10,
-            )
-        ]
-        prompt = build_mission_filter_prompt("Be a PM for engineering team", candidates)
-        assert "Be a PM for engineering team" in prompt
-
-    def test_prompt_contains_candidates(self):
-        """Test that prompt includes all candidates."""
-        candidates = [
-            EmergentCandidate(
-                name="Alice Chen",
-                detection_method="named_entity_extraction",
-                mention_count=10,
-            ),
-            EmergentCandidate(
-                name="Project Phoenix",
-                detection_method="named_entity_extraction",
-                mention_count=5,
-            ),
-        ]
-        prompt = build_mission_filter_prompt("Track projects", candidates)
-        assert "Alice Chen" in prompt
-        assert "Project Phoenix" in prompt
-
-    def test_prompt_contains_rejection_guidance(self):
-        """Test that prompt contains guidance to reject generic entities."""
-        candidates = [
-            EmergentCandidate(
-                name="test",
-                detection_method="named_entity_extraction",
-                mention_count=1,
-            )
-        ]
-        prompt = build_mission_filter_prompt("Test mission", candidates)
-
-        # Should contain rejection guidance for generic terms
-        assert "promote=false" in prompt
-        assert "kids" in prompt  # Example of generic term to reject
-        assert "community" in prompt  # Example of abstract concept to reject
-        assert "motivation" in prompt  # Example of abstract concept to reject
-
-
-class TestFilterCandidatesByMission:
-    """Test the filter_candidates_by_mission function."""
-
-    @pytest.fixture
-    def mock_llm_config(self):
-        """Create a mock LLM config."""
-        config = MagicMock()
-        config.call = AsyncMock()
-        return config
-
-    async def test_empty_candidates(self, mock_llm_config):
-        """Test with empty candidate list."""
-        result = await filter_candidates_by_mission(
-            llm_config=mock_llm_config,
-            mission="Test mission",
-            candidates=[],
-        )
-        assert result == []
-        mock_llm_config.call.assert_not_called()
-
-    async def test_no_mission_keeps_all(self, mock_llm_config):
-        """Test that no mission keeps all candidates (skips filtering)."""
-        candidates = [
-            EmergentCandidate(
-                name="Alice",
-                detection_method="named_entity_extraction",
-                mention_count=10,
-            )
-        ]
-        result = await filter_candidates_by_mission(
-            llm_config=mock_llm_config,
-            mission="",  # Empty mission
-            candidates=candidates,
-        )
-        assert len(result) == 1
-        assert result[0].name == "Alice"
-        mock_llm_config.call.assert_not_called()
-
-    async def test_filters_by_promote_flag(self, mock_llm_config):
-        """Test that candidates are filtered by promote flag."""
-        candidates = [
-            EmergentCandidate(
-                name="Alice Chen",
-                detection_method="named_entity_extraction",
-                mention_count=10,
-            ),
-            EmergentCandidate(
-                name="community",
-                detection_method="named_entity_extraction",
-                mention_count=5,
-            ),
-        ]
-
-        # Mock LLM response - Alice is promoted, community is not
-        mock_llm_config.call.return_value = MissionFilterResponse(
-            candidates=[
-                MissionFilterCandidate(name="Alice Chen", promote=True, reason="Specific person"),
-                MissionFilterCandidate(name="community", promote=False, reason="Generic abstract concept"),
-            ]
-        )
-
-        result = await filter_candidates_by_mission(
-            llm_config=mock_llm_config,
-            mission="Be a PM for engineering team",
-            candidates=candidates,
-        )
-
-        assert len(result) == 1
-        assert result[0].name == "Alice Chen"
-
-    async def test_rejects_generic_entities(self, mock_llm_config):
-        """Test that generic entities are rejected."""
-        # These are all generic/abstract terms that should be rejected
-        generic_names = [
-            "user", "support", "community", "family", "motivation",
-            "photo", "gratitude", "difference", "volunteering",
-            "kids", "veterans", "impact", "kindness", "encouragement",
-            "education", "nature", "joy", "positivity", "inspiration",
-            "help", "commitment", "passion", "energy", "connection",
-        ]
-        candidates = [
-            EmergentCandidate(
-                name=name,
-                detection_method="named_entity_extraction",
-                mention_count=10,
-            )
-            for name in generic_names
-        ]
-
-        # Add some valid candidates
-        valid_candidates = [
-            EmergentCandidate(
-                name="John",
-                detection_method="named_entity_extraction",
-                mention_count=10,
-            ),
-            EmergentCandidate(
-                name="Maria",
-                detection_method="named_entity_extraction",
-                mention_count=8,
-            ),
-            EmergentCandidate(
-                name="Max",
-                detection_method="named_entity_extraction",
-                mention_count=6,
-            ),
-        ]
-        candidates.extend(valid_candidates)
-
-        # Mock LLM response - reject all generic, promote only specific names
-        response_candidates = [
-            MissionFilterCandidate(name=name, promote=False, reason="Generic/abstract term")
-            for name in generic_names
-        ]
-        response_candidates.extend([
-            MissionFilterCandidate(name=c.name, promote=True, reason="Specific person name")
-            for c in valid_candidates
-        ])
-
-        mock_llm_config.call.return_value = MissionFilterResponse(candidates=response_candidates)
-
-        result = await filter_candidates_by_mission(
-            llm_config=mock_llm_config,
-            mission="Be a health coach",
-            candidates=candidates,
-        )
-
-        # Should only have John, Maria, and Max
-        result_names = {c.name for c in result}
-        assert result_names == {"John", "Maria", "Max"}
-
-    async def test_accepts_specific_named_entities(self, mock_llm_config):
-        """Test that specific named entities are accepted."""
-        # These should all be accepted
-        valid_names = [
-            "Alice Chen",       # Full name
-            "Dr. Smith",        # Title + name
-            "John",             # First name (when it's clearly a person)
-            "Google",           # Organization
-            "Frontend Team",    # Named team
-            "Project Phoenix",  # Named project
-            "NYC Office",       # Named place
-            "Q4 Planning",      # Named event
-            "Sprint 23 Review", # Named meeting
-        ]
-        candidates = [
-            EmergentCandidate(
-                name=name,
-                detection_method="named_entity_extraction",
-                mention_count=10,
-            )
-            for name in valid_names
-        ]
-
-        # Mock LLM response - promote all
-        response_candidates = [
-            MissionFilterCandidate(name=name, promote=True, reason="Specific named entity")
-            for name in valid_names
-        ]
-        mock_llm_config.call.return_value = MissionFilterResponse(candidates=response_candidates)
-
-        result = await filter_candidates_by_mission(
-            llm_config=mock_llm_config,
-            mission="Be a PM for engineering team",
-            candidates=candidates,
-        )
-
-        # Should have all valid names
-        result_names = {c.name for c in result}
-        assert result_names == set(valid_names)
-
-    async def test_llm_error_rejects_all_candidates(self, mock_llm_config):
-        """Test that LLM errors result in rejecting all candidates (fail-safe)."""
-        candidates = [
-            EmergentCandidate(
-                name="Alice",
-                detection_method="named_entity_extraction",
-                mention_count=10,
-            )
-        ]
-
-        mock_llm_config.call.side_effect = Exception("LLM error")
-
-        result = await filter_candidates_by_mission(
-            llm_config=mock_llm_config,
-            mission="Test mission",
-            candidates=candidates,
-        )
-
-        # Should reject all candidates on error (fail-safe)
-        assert len(result) == 0
-
-    async def test_missing_candidate_in_response_is_rejected(self, mock_llm_config):
-        """Test that candidates not in LLM response are rejected by default."""
-        candidates = [
-            EmergentCandidate(
-                name="Alice",
-                detection_method="named_entity_extraction",
-                mention_count=10,
-            ),
-            EmergentCandidate(
-                name="Bob",
-                detection_method="named_entity_extraction",
-                mention_count=5,
-            ),
-        ]
-
-        # Mock LLM response - only includes Alice, not Bob
-        mock_llm_config.call.return_value = MissionFilterResponse(
-            candidates=[
-                MissionFilterCandidate(name="Alice", promote=True, reason="Specific person"),
-            ]
-        )
-
-        result = await filter_candidates_by_mission(
-            llm_config=mock_llm_config,
-            mission="Test mission",
-            candidates=candidates,
-        )
-
-        # Only Alice should be in result (Bob was missing from response, so rejected)
-        assert len(result) == 1
-        assert result[0].name == "Alice"
-
-
-class TestEvaluateEmergentModels:
-    """Test the evaluate_emergent_models function for cleanup of existing models."""
-
-    @pytest.fixture
-    def mock_llm_config(self):
-        """Create a mock LLM config."""
-        config = MagicMock()
-        config.call = AsyncMock()
-        return config
-
-    async def test_empty_models(self, mock_llm_config):
-        """Test with empty model list."""
-        result = await evaluate_emergent_models(
-            llm_config=mock_llm_config,
-            models=[],
-        )
-        assert result == []
-        mock_llm_config.call.assert_not_called()
-
-    async def test_removes_generic_models(self, mock_llm_config):
-        """Test that generic/abstract models are marked for removal."""
-        models = [
-            {"id": "id-kids", "name": "kids"},
-            {"id": "id-community", "name": "community"},
-            {"id": "id-motivation", "name": "motivation"},
-            {"id": "id-john", "name": "John"},
-            {"id": "id-maria", "name": "Maria"},
-        ]
-
-        # Mock LLM response - reject generic, keep specific names
-        mock_llm_config.call.return_value = MissionFilterResponse(
-            candidates=[
-                MissionFilterCandidate(name="kids", promote=False, reason="Generic category"),
-                MissionFilterCandidate(name="community", promote=False, reason="Abstract concept"),
-                MissionFilterCandidate(name="motivation", promote=False, reason="Abstract concept"),
-                MissionFilterCandidate(name="John", promote=True, reason="Person name"),
-                MissionFilterCandidate(name="Maria", promote=True, reason="Person name"),
-            ]
-        )
-
-        result = await evaluate_emergent_models(
-            llm_config=mock_llm_config,
-            models=models,
-        )
-
-        # Should return IDs of generic models to remove
-        assert set(result) == {"id-kids", "id-community", "id-motivation"}
-
-    async def test_keeps_specific_named_models(self, mock_llm_config):
-        """Test that specific named models are kept."""
-        models = [
-            {"id": "id-john", "name": "John"},
-            {"id": "id-google", "name": "Google"},
-            {"id": "id-project", "name": "Project Phoenix"},
-        ]
-
-        # Mock LLM response - keep all
-        mock_llm_config.call.return_value = MissionFilterResponse(
-            candidates=[
-                MissionFilterCandidate(name="John", promote=True, reason="Person name"),
-                MissionFilterCandidate(name="Google", promote=True, reason="Organization"),
-                MissionFilterCandidate(name="Project Phoenix", promote=True, reason="Named project"),
-            ]
-        )
-
-        result = await evaluate_emergent_models(
-            llm_config=mock_llm_config,
-            models=models,
-        )
-
-        # No models should be removed
-        assert result == []
-
-    async def test_llm_error_keeps_all_models(self, mock_llm_config):
-        """Test that LLM errors result in keeping all models (safe default)."""
-        models = [
-            {"id": "id-kids", "name": "kids"},
-            {"id": "id-john", "name": "John"},
-        ]
-
-        mock_llm_config.call.side_effect = Exception("LLM error")
-
-        result = await evaluate_emergent_models(
-            llm_config=mock_llm_config,
-            models=models,
-        )
-
-        # Should keep all models on error (return empty removal list)
-        assert result == []
-
-    async def test_missing_model_in_response_is_removed(self, mock_llm_config):
-        """Test that models not in LLM response are marked for removal."""
-        models = [
-            {"id": "id-alice", "name": "Alice"},
-            {"id": "id-bob", "name": "Bob"},
-        ]
-
-        # Mock LLM response - only includes Alice
-        mock_llm_config.call.return_value = MissionFilterResponse(
-            candidates=[
-                MissionFilterCandidate(name="Alice", promote=True, reason="Person name"),
-            ]
-        )
-
-        result = await evaluate_emergent_models(
-            llm_config=mock_llm_config,
-            models=models,
-        )
-
-        # Bob should be marked for removal (missing from response)
-        assert result == ["id-bob"]
-
-
-class TestRemovedEntitiesNotRepromoted:
-    """Test that entities removed by evaluation are not re-promoted.
-
-    This tests the fix for a bug where:
-    1. evaluate_emergent_models returns model IDs to remove (e.g., 'entity-maya')
-    2. We delete those models
-    3. detect_entity_candidates finds the same entities (now eligible since model was deleted)
-    4. filter_candidates_by_goal approves them (different LLM call)
-    5. BUG: We were re-promoting the same entities we just removed
-
-    The fix tracks removed entity_ids and excludes them from promotion.
-    """
-
-    async def test_removed_entity_ids_excluded_from_promotion(self):
-        """Test that entities whose models were removed are not re-promoted."""
-        from hindsight_api.engine.mental_models.models import EmergentCandidate
-
-        # Simulate the scenario from the bug:
-        # - existing_emergent has model 'entity-maya' with entity_id='uuid-maya'
-        # - evaluate_emergent_models says to remove 'entity-maya'
-        # - detect_entity_candidates returns 'Maya' with entity_id='uuid-maya' (now eligible)
-        # - filter_candidates_by_goal says to promote 'Maya'
-        # - But we should NOT promote because we just removed it
-
-        existing_emergent = [
-            {"id": "entity-maya", "name": "Maya", "entity_id": "uuid-maya"},
-            {"id": "entity-alex", "name": "Alex", "entity_id": "uuid-alex"},
-            {"id": "entity-john", "name": "John", "entity_id": "uuid-john"},  # This one will be kept
-        ]
-
-        # Models to remove (evaluate_emergent_models would return these)
-        models_to_remove = ["entity-maya", "entity-alex"]
-
-        # Build model_id -> entity_id mapping (this is what the fix does)
-        model_to_entity = {m["id"]: m.get("entity_id") for m in existing_emergent}
-
-        # Track removed entity_ids
-        removed_entity_ids: set[str] = set()
-        for model_id in models_to_remove:
-            entity_id = model_to_entity.get(model_id)
-            if entity_id:
-                removed_entity_ids.add(str(entity_id))
-
-        # Verify we tracked the right entity_ids
-        assert removed_entity_ids == {"uuid-maya", "uuid-alex"}
-
-        # Now simulate candidates that were detected (includes removed entities)
-        candidates = [
-            EmergentCandidate(
-                name="Maya", entity_id="uuid-maya", detection_method="named_entity", mention_count=10
-            ),
-            EmergentCandidate(
-                name="Alex", entity_id="uuid-alex", detection_method="named_entity", mention_count=8
-            ),
-            EmergentCandidate(
-                name="NewPerson", entity_id="uuid-new", detection_method="named_entity", mention_count=5
-            ),
-        ]
-
-        # Filter out candidates whose entity was just removed (the fix)
-        filtered_candidates = [c for c in candidates if c.entity_id not in removed_entity_ids]
-
-        # Only NewPerson should remain - Maya and Alex were removed and should not be re-promoted
-        assert len(filtered_candidates) == 1
-        assert filtered_candidates[0].name == "NewPerson"
-        assert filtered_candidates[0].entity_id == "uuid-new"
-
-    async def test_candidates_without_matching_removal_are_kept(self):
-        """Test that candidates not in the removed set are still promoted."""
-        from hindsight_api.engine.mental_models.models import EmergentCandidate
-
-        # No models removed
-        removed_entity_ids: set[str] = set()
-
-        candidates = [
-            EmergentCandidate(
-                name="Alice", entity_id="uuid-alice", detection_method="named_entity", mention_count=10
-            ),
-            EmergentCandidate(
-                name="Bob", entity_id="uuid-bob", detection_method="named_entity", mention_count=8
-            ),
-        ]
-
-        # Filter (should keep all since nothing was removed)
-        filtered_candidates = [c for c in candidates if c.entity_id not in removed_entity_ids]
-
-        assert len(filtered_candidates) == 2
-        assert {c.name for c in filtered_candidates} == {"Alice", "Bob"}
-
-    async def test_partial_removal_keeps_other_candidates(self):
-        """Test that only removed entities are excluded, others pass through."""
-        from hindsight_api.engine.mental_models.models import EmergentCandidate
-
-        # Only one entity removed
-        removed_entity_ids = {"uuid-removed"}
-
-        candidates = [
-            EmergentCandidate(
-                name="Removed", entity_id="uuid-removed", detection_method="named_entity", mention_count=10
-            ),
-            EmergentCandidate(
-                name="Kept1", entity_id="uuid-kept1", detection_method="named_entity", mention_count=8
-            ),
-            EmergentCandidate(
-                name="Kept2", entity_id="uuid-kept2", detection_method="named_entity", mention_count=5
-            ),
-        ]
-
-        filtered_candidates = [c for c in candidates if c.entity_id not in removed_entity_ids]
-
-        assert len(filtered_candidates) == 2
-        assert {c.name for c in filtered_candidates} == {"Kept1", "Kept2"}

hindsight-api/tests/test_extensions.py

@@ -17,8 +17,6 @@ from hindsight_api.extensions import (
     RecallResult,
     ReflectContext,
     ReflectResultContext,
-    RefreshMentalModelContext,
-    RefreshMentalModelResult,
     RequestContext,
     RetainContext,
     RetainResult,
@@ -95,7 +93,6 @@ class RateLimitingValidator(OperationValidatorExtension):
         self.retain_counts: dict[str, int] = defaultdict(int)
         self.recall_counts: dict[str, int] = defaultdict(int)
         self.reflect_counts: dict[str, int] = defaultdict(int)
-        self.refresh_mental_model_counts: dict[str, int] = defaultdict(int)
 
     async def validate_retain(self, ctx: RetainContext) -> ValidationResult:
         self.retain_counts[ctx.bank_id] += 1
@@ -121,16 +118,6 @@ class RateLimitingValidator(OperationValidatorExtension):
             )
         return ValidationResult.accept()
 
-    async def validate_refresh_mental_model(
-        self, ctx: RefreshMentalModelContext
-    ) -> ValidationResult:
-        self.refresh_mental_model_counts[ctx.bank_id] += 1
-        if self.refresh_mental_model_counts[ctx.bank_id] > self.max_attempts:
-            return ValidationResult.reject(
-                f"Refresh mental model limit exceeded for bank {ctx.bank_id}"
-            )
-        return ValidationResult.accept()
-
 
 class TrackingValidator(OperationValidatorExtension):
     """
@@ -145,12 +132,10 @@ class TrackingValidator(OperationValidatorExtension):
         self.pre_retain_calls: list[RetainContext] = []
         self.pre_recall_calls: list[RecallContext] = []
         self.pre_reflect_calls: list[ReflectContext] = []
-        self.pre_refresh_mental_model_calls: list[RefreshMentalModelContext] = []
         # Post-hook tracking
         self.post_retain_calls: list[RetainResult] = []
         self.post_recall_calls: list[RecallResult] = []
         self.post_reflect_calls: list[ReflectResultContext] = []
-        self.post_refresh_mental_model_calls: list[RefreshMentalModelResult] = []
 
     async def validate_retain(self, ctx: RetainContext) -> ValidationResult:
         self.pre_retain_calls.append(ctx)
@@ -164,12 +149,6 @@ class TrackingValidator(OperationValidatorExtension):
         self.pre_reflect_calls.append(ctx)
         return ValidationResult.accept()
 
-    async def validate_refresh_mental_model(
-        self, ctx: RefreshMentalModelContext
-    ) -> ValidationResult:
-        self.pre_refresh_mental_model_calls.append(ctx)
-        return ValidationResult.accept()
-
     async def on_retain_complete(self, result: RetainResult) -> None:
         self.post_retain_calls.append(result)
 
@@ -179,11 +158,6 @@ class TrackingValidator(OperationValidatorExtension):
     async def on_reflect_complete(self, result: ReflectResultContext) -> None:
         self.post_reflect_calls.append(result)
 
-    async def on_refresh_mental_model_complete(
-        self, result: RefreshMentalModelResult
-    ) -> None:
-        self.post_refresh_mental_model_calls.append(result)
-
 
 class TestMemoryEngineValidation:
     """Tests for validation integration with MemoryEngine.
@@ -541,105 +515,6 @@ class TestOperationHooksParameters:
         assert len(validator.pre_recall_calls) == 1
         assert len(validator.post_recall_calls) == 1
 
-    @pytest.mark.asyncio
-    async def test_refresh_mental_model_pre_hook_receives_all_parameters(
-        self, memory_with_tracking_validator
-    ):
-        """Pre-refresh-mental-model hook receives all user-provided parameters."""
-        import uuid
-
-        memory, validator = memory_with_tracking_validator
-        bank_id = f"test-refresh-mm-params-{uuid.uuid4().hex[:8]}"
-        ctx = RequestContext(api_key="test-key")
-
-        # Create bank first (get_bank_profile auto-creates if needed)
-        await memory.get_bank_profile(bank_id, request_context=ctx)
-
-        # Create a pinned mental model
-        model = await memory.create_mental_model(
-            bank_id=bank_id,
-            name="Test Model",
-            description="Test description",
-            subtype="pinned",
-            request_context=ctx,
-        )
-
-        assert model is not None
-        model_id = model["id"]
-
-        # Attempt to refresh (may not actually refresh if no data, but hook should be called)
-        try:
-            await memory.refresh_mental_model(
-                bank_id=bank_id,
-                model_id=model_id,
-                request_context=ctx,
-            )
-        except Exception:
-            pass  # May fail if no data
-
-        # Check pre-hook was called
-        assert len(validator.pre_refresh_mental_model_calls) == 1
-        pre_ctx = validator.pre_refresh_mental_model_calls[0]
-        assert pre_ctx.bank_id == bank_id
-        assert pre_ctx.model_id == model_id
-        assert pre_ctx.request_context == ctx
-
-    @pytest.mark.asyncio
-    async def test_refresh_mental_model_post_hook_receives_token_usage(
-        self, memory_with_tracking_validator
-    ):
-        """Post-refresh-mental-model hook receives token usage information."""
-        import uuid
-
-        memory, validator = memory_with_tracking_validator
-        bank_id = f"test-refresh-mm-tokens-{uuid.uuid4().hex[:8]}"
-        ctx = RequestContext(api_key="test-key")
-
-        # Store some content first
-        await memory.retain_batch_async(
-            bank_id=bank_id,
-            contents=[
-                {"content": "Alice is a software engineer who works on machine learning."},
-                {"content": "Alice enjoys hiking and outdoor activities on weekends."},
-                {"content": "Alice has been working at the company for 5 years."},
-            ],
-            request_context=ctx,
-        )
-
-        # Create a pinned mental model
-        model = await memory.create_mental_model(
-            bank_id=bank_id,
-            name="Alice Profile",
-            description="Profile of Alice including work and hobbies",
-            subtype="pinned",
-            request_context=ctx,
-        )
-
-        if model:
-            model_id = model["id"]
-
-            # Refresh the mental model
-            result = await memory.refresh_mental_model(
-                bank_id=bank_id,
-                model_id=model_id,
-                request_context=ctx,
-            )
-
-            # Check post-hook was called with token usage
-            if validator.post_refresh_mental_model_calls:
-                post_result = validator.post_refresh_mental_model_calls[0]
-                assert post_result.bank_id == bank_id
-                assert post_result.model_id == model_id
-                assert post_result.request_context == ctx
-                assert post_result.success is True
-                assert post_result.error is None
-
-                # Token usage should be populated (may be 0 if refresh was skipped)
-                assert post_result.total_tokens >= 0
-                assert post_result.input_tokens >= 0
-                assert post_result.output_tokens >= 0
-                assert post_result.duration_ms >= 0
-
 
 class TestTenantExtension:
     """Tests for TenantExtension and ApiKeyTenantExtension."""

hindsight-api/tests/test_llm_tools.py

@@ -241,24 +241,27 @@ class TestReflectToolSchemas:
         tools = get_reflect_tools()
 
         tool_names = [t["function"]["name"] for t in tools]
-        assert "list_mental_models" in tool_names
-        assert "get_mental_model" in tool_names
+        assert "search_reflections" in tool_names
+        assert "search_mental_models" in tool_names
         assert "recall" in tool_names
-        assert "learn" in tool_names
         assert "expand" in tool_names
         assert "done" in tool_names
 
-    def test_get_reflect_tools_without_learn(self):
-        """Test getting reflect tools without learn."""
+    def test_get_reflect_tools_with_directives(self):
+        """Test getting reflect tools with directive rules."""
         from hindsight_api.engine.reflect.tools_schema import get_reflect_tools
 
-        tools = get_reflect_tools(enable_learn=False)
+        tools = get_reflect_tools(directive_rules=["Always respond in French"])
 
         tool_names = [t["function"]["name"] for t in tools]
-        assert "learn" not in tool_names
         assert "recall" in tool_names
         assert "done" in tool_names
 
+        # Done tool should have directive_compliance field when directives are present
+        done_tool = next(t for t in tools if t["function"]["name"] == "done")
+        params = done_tool["function"]["parameters"]["properties"]
+        assert "directive_compliance" in params
+
     def test_get_reflect_tools_answer_mode(self):
         """Test getting reflect tools with answer output mode."""
         from hindsight_api.engine.reflect.tools_schema import get_reflect_tools
@@ -270,7 +273,8 @@ class TestReflectToolSchemas:
 
         assert "answer" in params
         assert "memory_ids" in params
-        assert "model_ids" in params
+        assert "mental_model_ids" in params
+        assert "reflection_ids" in params
 
 
 class TestLLMToolCallResult:

hindsight-api/tests/test_load_large_batch.py

@@ -19,6 +19,7 @@ import pytest_asyncio
 from hindsight_api import MemoryEngine, LLMConfig, LocalSTEmbeddings, RequestContext
 from hindsight_api.engine.cross_encoder import LocalSTCrossEncoder
 from hindsight_api.engine.query_analyzer import DateparserQueryAnalyzer
+from hindsight_api.engine.task_backend import SyncTaskBackend
 from hindsight_api.engine.retain.fact_extraction import FactExtractionResponse, ExtractedFact
 from hindsight_api.engine.llm_wrapper import TokenUsage
 
@@ -106,6 +107,7 @@ class TestLargeBatchRetain:
             pool_max_size=10,
             run_migrations=False,
             skip_llm_verification=True,  # Skip LLM verification since we're mocking
+            task_backend=SyncTaskBackend(),  # Execute tasks immediately in tests
         )
         await mem.initialize()
         yield mem

hindsight-api/tests/test_main_module.py

@@ -363,7 +363,6 @@ from hindsight_api.extensions import (
     RetainContext,
     RecallContext,
     ReflectContext,
-    RefreshMentalModelContext,
 )
 
 
@@ -395,6 +394,3 @@ class MockOperationValidator(OperationValidatorExtension):
 
     async def validate_reflect(self, ctx: ReflectContext) -> ValidationResult:
         return ValidationResult.accept()
-
-    async def validate_refresh_mental_model(self, ctx: RefreshMentalModelContext) -> ValidationResult:
-        return ValidationResult.accept()

hindsight-api/tests/test_multilingual.py

@@ -275,6 +275,165 @@ async def test_retain_japanese_content(memory, request_context):
                 pass
 
 
+@pytest.mark.asyncio
+async def test_english_content_stays_english(memory, request_context):
+    """
+    Test that English content is NOT incorrectly translated to Japanese or Chinese.
+
+    This test specifically catches the bug where the language instruction in the
+    CONCISE extraction prompt mentioned Japanese/Chinese explicitly, which primed
+    the LLM to sometimes output facts in those languages even for English input.
+
+    See: https://github.com/vectorize-io/hindsight/issues/181
+    """
+    bank_id = f"test_english_retain_{datetime.now(timezone.utc).timestamp()}"
+
+    try:
+        # English content about a developer
+        english_content = """
+        John Smith is a software engineer at TechCorp in Seattle.
+        He specializes in machine learning and has been working on
+        recommendation systems for the past three years.
+        Last month, he launched a new feature that improved click-through rates by 25%.
+        He prefers working in Python and uses PyTorch for model training.
+        """
+
+        unit_ids = await memory.retain_async(
+            bank_id=bank_id,
+            content=english_content,
+            context="Team profile",
+            event_date=datetime(2024, 1, 15, tzinfo=timezone.utc),
+            request_context=request_context,
+        )
+
+        logger.info(f"Retained {len(unit_ids)} facts from English content")
+        assert len(unit_ids) > 0, "Should have extracted facts from English content"
+
+        # Recall with English query
+        result = await memory.recall_async(
+            bank_id=bank_id,
+            query="Tell me about John Smith",
+            budget=Budget.MID,
+            max_tokens=1000,
+            fact_type=["world"],
+            request_context=request_context,
+        )
+
+        assert len(result.results) > 0, "Should recall facts about John Smith"
+
+        # Verify facts are NOT in Japanese or Chinese
+        for fact in result.results:
+            logger.info(f"Fact: {fact.text}")
+
+            # Count Japanese characters (hiragana, katakana)
+            japanese_chars = sum(
+                1 for char in fact.text
+                if ("\u3040" <= char <= "\u309f") or ("\u30a0" <= char <= "\u30ff")
+            )
+
+            # Count Chinese/CJK characters (excluding those also used in Japanese)
+            # Note: Kanji/CJK ideographs overlap between Chinese and Japanese
+            cjk_chars = sum(1 for char in fact.text if "\u4e00" <= char <= "\u9fff")
+
+            # For English input, there should be minimal CJK characters
+            # Allow for occasional edge cases (e.g., proper nouns) but not full translation
+            total_chars = len(fact.text)
+            cjk_ratio = cjk_chars / max(total_chars, 1)
+
+            assert cjk_ratio < 0.1, (
+                f"English content was incorrectly translated to CJK language! "
+                f"CJK ratio: {cjk_ratio:.1%}, Japanese chars: {japanese_chars}, CJK chars: {cjk_chars}. "
+                f"Fact: {fact.text}"
+            )
+
+        logger.info("English content test passed - facts stayed in English")
+
+    finally:
+        await memory.delete_bank(bank_id, request_context=request_context)
+
+
+@pytest.mark.asyncio
+async def test_italian_content_stays_italian(memory, request_context):
+    """
+    Test that Italian content is NOT incorrectly translated to Japanese or Chinese.
+
+    Similar to the English test, this catches the bug where non-CJK languages
+    could be incorrectly translated due to biased language instruction.
+
+    See: https://github.com/vectorize-io/hindsight/issues/181
+    """
+    bank_id = f"test_italian_retain_{datetime.now(timezone.utc).timestamp()}"
+
+    try:
+        # Italian content about a chef
+        italian_content = """
+        Marco Rossi è uno chef italiano che lavora in un ristorante a Milano.
+        È specializzato nella cucina toscana e ha vinto tre premi gastronomici.
+        Il mese scorso ha aperto un nuovo ristorante nel centro della città.
+        Preferisce usare ingredienti freschi e locali per i suoi piatti.
+        """
+
+        unit_ids = await memory.retain_async(
+            bank_id=bank_id,
+            content=italian_content,
+            context="Profilo dello chef",
+            event_date=datetime(2024, 1, 15, tzinfo=timezone.utc),
+            request_context=request_context,
+        )
+
+        logger.info(f"Retained {len(unit_ids)} facts from Italian content")
+        assert len(unit_ids) > 0, "Should have extracted facts from Italian content"
+
+        # Recall with Italian query
+        result = await memory.recall_async(
+            bank_id=bank_id,
+            query="Dimmi di Marco Rossi",  # "Tell me about Marco Rossi"
+            budget=Budget.MID,
+            max_tokens=1000,
+            fact_type=["world"],
+            request_context=request_context,
+        )
+
+        assert len(result.results) > 0, "Should recall facts about Marco Rossi"
+
+        # Verify facts are NOT in Japanese or Chinese - should stay in Italian
+        for fact in result.results:
+            logger.info(f"Fact: {fact.text}")
+
+            # Count CJK characters
+            cjk_chars = sum(1 for char in fact.text if "\u4e00" <= char <= "\u9fff")
+            japanese_chars = sum(
+                1 for char in fact.text
+                if ("\u3040" <= char <= "\u309f") or ("\u30a0" <= char <= "\u30ff")
+            )
+
+            total_chars = len(fact.text)
+            cjk_ratio = (cjk_chars + japanese_chars) / max(total_chars, 1)
+
+            assert cjk_ratio < 0.1, (
+                f"Italian content was incorrectly translated to CJK language! "
+                f"CJK ratio: {cjk_ratio:.1%}. Fact: {fact.text}"
+            )
+
+        # Verify facts contain Italian words (basic sanity check)
+        all_text = " ".join(f.text for f in result.results).lower()
+        italian_indicators = ["marco", "rossi", "chef", "ristorante", "milano", "cucina", "italiano", "italiana"]
+        has_italian = any(word in all_text for word in italian_indicators)
+
+        # Allow English translation as acceptable (not ideal but not the bug)
+        english_indicators = ["chef", "restaurant", "milan", "italian", "cooking"]
+        has_english = any(word in all_text for word in english_indicators)
+
+        assert has_italian or has_english, (
+            f"Expected facts to be in Italian or English, but got neither. Facts: {all_text}"
+        )
+
+        logger.info("Italian content test passed - facts not translated to CJK")
+
+    finally:
+        await memory.delete_bank(bank_id, request_context=request_context)
+
+
 @pytest.mark.asyncio
 async def test_mixed_language_entities(memory, request_context):
     """

hindsight-api/tests/test_observation_trends.py

@@ -1,405 +0,0 @@
-"""Tests for observation trend computation and evidence-grounded models."""
-
-from datetime import datetime, timedelta, timezone
-
-import pytest
-
-from hindsight_api.engine.reflect.observations import (
-    CandidateObservation,
-    Observation,
-    ObservationEvidence,
-    Trend,
-    compute_trend,
-    verify_evidence_quotes,
-)
-
-
-class TestComputeTrend:
-    """Tests for the compute_trend function."""
-
-    def test_empty_evidence_returns_stale(self):
-        """No evidence should return STALE trend."""
-        trend = compute_trend([])
-        assert trend == Trend.STALE
-
-    def test_all_recent_evidence_returns_new(self):
-        """All evidence within recent window (30 days) should return NEW trend.
-
-        Scenario: User just started using the app and mentioned they like coffee twice.
-        Both mentions are within the last 2 weeks, so this is a NEW observation.
-        """
-        now = datetime.now(timezone.utc)
-        evidence = [
-            ObservationEvidence(
-                memory_id="mem-coffee-morning",
-                quote="I always start my day with a large black coffee",
-                relevance="Shows preference for coffee and morning routine",
-                timestamp=now - timedelta(days=5),
-            ),
-            ObservationEvidence(
-                memory_id="mem-coffee-meeting",
-                quote="grabbed coffee before the standup meeting",
-                relevance="Confirms regular coffee consumption",
-                timestamp=now - timedelta(days=10),
-            ),
-        ]
-
-        trend = compute_trend(evidence, now=now)
-        assert trend == Trend.NEW
-
-    def test_no_recent_evidence_returns_stale(self):
-        """No evidence in recent window should return STALE trend.
-
-        Scenario: User mentioned running 3 months ago but hasn't mentioned it since.
-        The observation about running as a hobby may no longer be accurate.
-        """
-        now = datetime.now(timezone.utc)
-        evidence = [
-            ObservationEvidence(
-                memory_id="mem-running-march",
-                quote="training for a half marathon in the spring",
-                relevance="Shows interest in running",
-                timestamp=now - timedelta(days=60),
-            ),
-            ObservationEvidence(
-                memory_id="mem-running-feb",
-                quote="went for a 10k run this morning",
-                relevance="Active runner",
-                timestamp=now - timedelta(days=100),
-            ),
-        ]
-
-        trend = compute_trend(evidence, now=now)
-        assert trend == Trend.STALE
-
-    def test_stable_evidence_distribution(self):
-        """Evidence spread evenly across time should return STABLE trend.
-
-        Scenario: User has consistently mentioned working remotely over 4 months.
-        Evidence is well-distributed, indicating a stable, ongoing preference.
-        """
-        now = datetime.now(timezone.utc)
-        evidence = [
-            # Recent (within 30 days)
-            ObservationEvidence(
-                memory_id="mem-remote-jan",
-                quote="working from my home office today",
-                relevance="Current remote work",
-                timestamp=now - timedelta(days=5),
-            ),
-            ObservationEvidence(
-                memory_id="mem-remote-dec",
-                quote="the flexibility of remote work is great",
-                relevance="Values remote work",
-                timestamp=now - timedelta(days=15),
-            ),
-            # Middle period (30-90 days)
-            ObservationEvidence(
-                memory_id="mem-remote-nov",
-                quote="set up a standing desk at home",
-                relevance="Invested in home office",
-                timestamp=now - timedelta(days=45),
-            ),
-            ObservationEvidence(
-                memory_id="mem-remote-oct",
-                quote="prefer async communication over meetings",
-                relevance="Remote work style preference",
-                timestamp=now - timedelta(days=60),
-            ),
-            # Older (90+ days)
-            ObservationEvidence(
-                memory_id="mem-remote-sep",
-                quote="switched to fully remote last quarter",
-                relevance="Original transition to remote",
-                timestamp=now - timedelta(days=100),
-            ),
-            ObservationEvidence(
-                memory_id="mem-remote-aug",
-                quote="negotiated remote work in my new contract",
-                relevance="Intentional choice for remote",
-                timestamp=now - timedelta(days=120),
-            ),
-        ]
-
-        trend = compute_trend(evidence, now=now)
-        assert trend == Trend.STABLE
-
-    def test_strengthening_trend(self):
-        """Much more recent evidence than older should return STRENGTHENING trend.
-
-        Scenario: User has been increasingly talking about learning Python recently
-        after mentioning it once months ago. Interest appears to be growing.
-        """
-        now = datetime.now(timezone.utc)
-        evidence = [
-            # Lots of recent evidence - actively learning
-            ObservationEvidence(
-                memory_id="mem-python-project",
-                quote="finished my first Python project - a web scraper",
-                relevance="Completed Python project",
-                timestamp=now - timedelta(days=2),
-            ),
-            ObservationEvidence(
-                memory_id="mem-python-course",
-                quote="halfway through the Python bootcamp",
-                relevance="Active learning",
-                timestamp=now - timedelta(days=5),
-            ),
-            ObservationEvidence(
-                memory_id="mem-python-book",
-                quote="reading Fluent Python, it's excellent",
-                relevance="Deepening knowledge",
-                timestamp=now - timedelta(days=10),
-            ),
-            ObservationEvidence(
-                memory_id="mem-python-practice",
-                quote="solved 50 LeetCode problems in Python",
-                relevance="Practicing skills",
-                timestamp=now - timedelta(days=15),
-            ),
-            ObservationEvidence(
-                memory_id="mem-python-ide",
-                quote="set up VS Code with all the Python extensions",
-                relevance="Setting up environment",
-                timestamp=now - timedelta(days=20),
-            ),
-            # Only one old mention - initial interest
-            ObservationEvidence(
-                memory_id="mem-python-start",
-                quote="thinking about learning Python someday",
-                relevance="Initial interest",
-                timestamp=now - timedelta(days=100),
-            ),
-        ]
-
-        trend = compute_trend(evidence, now=now)
-        assert trend == Trend.STRENGTHENING
-
-    def test_weakening_trend(self):
-        """Much less recent evidence than older should return WEAKENING trend.
-
-        Scenario: User was very active in a book club last year but mentions
-        have tapered off. The observation about being a book club member
-        may be becoming less relevant.
-        """
-        now = datetime.now(timezone.utc)
-        evidence = [
-            # Only one recent mention
-            ObservationEvidence(
-                memory_id="mem-book-recent",
-                quote="haven't had time for book club lately",
-                relevance="Reduced participation",
-                timestamp=now - timedelta(days=10),
-            ),
-            # Lots of older evidence - was very active
-            ObservationEvidence(
-                memory_id="mem-book-aug",
-                quote="hosting book club at my place next week",
-                relevance="Active organizer",
-                timestamp=now - timedelta(days=40),
-            ),
-            ObservationEvidence(
-                memory_id="mem-book-july",
-                quote="leading the discussion on 1984",
-                relevance="Active participant",
-                timestamp=now - timedelta(days=50),
-            ),
-            ObservationEvidence(
-                memory_id="mem-book-june",
-                quote="we picked The Midnight Library for June",
-                relevance="Regular member",
-                timestamp=now - timedelta(days=60),
-            ),
-            ObservationEvidence(
-                memory_id="mem-book-may",
-                quote="book club was amazing tonight",
-                relevance="Enthusiastic member",
-                timestamp=now - timedelta(days=100),
-            ),
-            ObservationEvidence(
-                memory_id="mem-book-april",
-                quote="joined a new book club in my neighborhood",
-                relevance="Started participation",
-                timestamp=now - timedelta(days=110),
-            ),
-            ObservationEvidence(
-                memory_id="mem-book-march",
-                quote="excited to finally join a book club",
-                relevance="Initial enthusiasm",
-                timestamp=now - timedelta(days=120),
-            ),
-        ]
-
-        trend = compute_trend(evidence, now=now)
-        assert trend == Trend.WEAKENING
-
-
-class TestObservationModel:
-    """Tests for the Observation model."""
-
-    def test_observation_computed_trend(self):
-        """Observation should have computed trend property based on evidence."""
-        now = datetime.now(timezone.utc)
-        obs = Observation(
-            title="Morning meeting preference",
-            content="Prefers morning meetings over afternoon ones",
-            evidence=[
-                ObservationEvidence(
-                    memory_id="mem-morning-standup",
-                    quote="I'm most productive in morning meetings",
-                    relevance="Direct preference statement",
-                    timestamp=now - timedelta(days=5),
-                ),
-            ],
-            created_at=now,
-        )
-
-        assert obs.trend == Trend.NEW
-        assert obs.evidence_count == 1
-
-    def test_observation_evidence_span(self):
-        """Observation should compute evidence span correctly.
-
-        The span shows the date range of supporting evidence, helping
-        understand how long this pattern has been observed.
-        """
-        now = datetime.now(timezone.utc)
-        old_time = now - timedelta(days=100)
-        recent_time = now - timedelta(days=5)
-
-        obs = Observation(
-            title="Values work-life balance",
-            content="Values work-life balance highly",
-            evidence=[
-                ObservationEvidence(
-                    memory_id="mem-balance-old",
-                    quote="turned down a promotion because of the hours",
-                    relevance="Prioritized balance over advancement",
-                    timestamp=old_time,
-                ),
-                ObservationEvidence(
-                    memory_id="mem-balance-recent",
-                    quote="always log off by 6pm no matter what",
-                    relevance="Maintains boundaries",
-                    timestamp=recent_time,
-                ),
-            ],
-            created_at=now,
-        )
-
-        evidence_span = obs.evidence_span
-        assert evidence_span["from"] == old_time.isoformat()
-        assert evidence_span["to"] == recent_time.isoformat()
-
-    def test_observation_empty_evidence_span(self):
-        """Observation with no evidence should have null span."""
-        obs = Observation(
-            title="Test observation",
-            content="Test observation without evidence",
-            evidence=[],
-        )
-
-        evidence_span = obs.evidence_span
-        assert evidence_span["from"] is None
-        assert evidence_span["to"] is None
-
-
-class TestVerifyEvidenceQuotes:
-    """Tests for evidence quote verification.
-
-    This ensures the LLM isn't hallucinating quotes - every quote
-    must actually appear in the source memory.
-    """
-
-    def test_valid_quotes(self):
-        """Should return True when quotes exist in their source memories."""
-        obs = Observation(
-            title="Enjoys hiking",
-            content="Enjoys hiking on weekends",
-            evidence=[
-                ObservationEvidence(
-                    memory_id="mem-hiking-trip",
-                    quote="went hiking at Mount Tam",
-                    relevance="Shows hiking activity",
-                    timestamp=datetime.now(timezone.utc),
-                ),
-            ],
-        )
-
-        memories = {
-            "mem-hiking-trip": "Had a great Saturday - went hiking at Mount Tam with friends and saw amazing views."
-        }
-        is_valid, errors = verify_evidence_quotes(obs, memories)
-
-        assert is_valid is True
-        assert len(errors) == 0
-
-    def test_invalid_quote(self):
-        """Should return False when quote doesn't exist in memory.
-
-        This catches LLM hallucinations where it fabricates quotes.
-        """
-        obs = Observation(
-            title="Loves spicy food",
-            content="Loves spicy food",
-            evidence=[
-                ObservationEvidence(
-                    memory_id="mem-dinner",
-                    quote="I love extra hot salsa",
-                    relevance="Shows spicy food preference",
-                    timestamp=datetime.now(timezone.utc),
-                ),
-            ],
-        )
-
-        memories = {"mem-dinner": "Had tacos for dinner. The guacamole was really fresh."}
-        is_valid, errors = verify_evidence_quotes(obs, memories)
-
-        assert is_valid is False
-        assert len(errors) == 1
-        assert "Quote not found" in errors[0]
-
-    def test_missing_memory(self):
-        """Should return False when referenced memory doesn't exist.
-
-        This catches cases where the LLM references a memory ID that
-        was never actually retrieved.
-        """
-        obs = Observation(
-            title="Has a dog named Max",
-            content="Has a dog named Max",
-            evidence=[
-                ObservationEvidence(
-                    memory_id="mem-pet-story",
-                    quote="took Max to the vet",
-                    relevance="Shows pet ownership",
-                    timestamp=datetime.now(timezone.utc),
-                ),
-            ],
-        )
-
-        memories = {"mem-different-id": "Some unrelated memory content"}
-        is_valid, errors = verify_evidence_quotes(obs, memories)
-
-        assert is_valid is False
-        assert len(errors) == 1
-        assert "not found" in errors[0]
-
-
-class TestCandidateObservation:
-    """Tests for candidate observation model.
-
-    Candidates are generated in the SEED phase and validated
-    before becoming full observations.
-    """
-
-    def test_create_candidate(self):
-        """Should create candidate with content and seed memories."""
-        candidate = CandidateObservation(
-            content="User prefers async communication over meetings",
-            seed_memory_ids=["mem-slack-pref", "mem-meeting-decline"],
-        )
-
-        assert candidate.content == "User prefers async communication over meetings"
-        assert len(candidate.seed_memory_ids) == 2
-        assert "mem-slack-pref" in candidate.seed_memory_ids

hindsight-api/tests/test_reflections.py

@@ -0,0 +1,359 @@
+"""Tests for reflections, mental models, and learnings functionality."""
+
+import uuid
+
+import pytest
+import pytest_asyncio
+import httpx
+from hindsight_api.api import create_app
+from hindsight_api.engine.memory_engine import MemoryEngine
+
+
+@pytest_asyncio.fixture
+async def api_client(memory):
+    """Create an async test client for the FastAPI app."""
+    app = create_app(memory, initialize_memory=False)
+    transport = httpx.ASGITransport(app=app)
+    async with httpx.AsyncClient(transport=transport, base_url="http://test") as client:
+        yield client
+
+
+@pytest.fixture
+def test_bank_id():
+    """Provide a unique bank ID for this test run."""
+    return f"test_reflections_{uuid.uuid4().hex[:8]}"
+
+
+class TestReflectionsCRUD:
+    """Test reflections CRUD operations via memory engine."""
+
+    @pytest.mark.asyncio
+    async def test_create_and_get_reflection(self, memory: MemoryEngine, request_context):
+        """Test creating and retrieving a reflection."""
+        bank_id = f"test-reflection-{uuid.uuid4().hex[:8]}"
+
+        # Create the bank first
+        await memory.get_bank_profile(bank_id=bank_id, request_context=request_context)
+
+        # Create a reflection
+        reflection = await memory.create_reflection(
+            bank_id=bank_id,
+            name="Team Preferences",
+            source_query="What are the team's communication preferences?",
+            content="The team prefers async communication via Slack",
+            tags=["team"],
+            request_context=request_context,
+        )
+
+        assert reflection["name"] == "Team Preferences"
+        assert reflection["source_query"] == "What are the team's communication preferences?"
+        assert reflection["content"] == "The team prefers async communication via Slack"
+        assert reflection["tags"] == ["team"]
+        assert "id" in reflection
+
+        # Get the reflection
+        fetched = await memory.get_reflection(
+            bank_id=bank_id,
+            reflection_id=reflection["id"],
+            request_context=request_context,
+        )
+
+        assert fetched["id"] == reflection["id"]
+        assert fetched["name"] == "Team Preferences"
+
+        # Cleanup
+        await memory.delete_bank(bank_id, request_context=request_context)
+
+    @pytest.mark.asyncio
+    async def test_list_reflections(self, memory: MemoryEngine, request_context):
+        """Test listing reflections with filters."""
+        bank_id = f"test-reflection-list-{uuid.uuid4().hex[:8]}"
+
+        # Create the bank first
+        await memory.get_bank_profile(bank_id=bank_id, request_context=request_context)
+
+        # Create multiple reflections
+        await memory.create_reflection(
+            bank_id=bank_id,
+            name="Reflection 1",
+            source_query="Query 1",
+            content="Content 1",
+            tags=["tag1"],
+            request_context=request_context,
+        )
+        await memory.create_reflection(
+            bank_id=bank_id,
+            name="Reflection 2",
+            source_query="Query 2",
+            content="Content 2",
+            tags=["tag2"],
+            request_context=request_context,
+        )
+
+        # List all
+        all_reflections = await memory.list_reflections(
+            bank_id=bank_id,
+            request_context=request_context,
+        )
+        assert len(all_reflections) == 2
+
+        # List with tag filter
+        tag1_reflections = await memory.list_reflections(
+            bank_id=bank_id,
+            tags=["tag1"],
+            request_context=request_context,
+        )
+        assert len(tag1_reflections) == 1
+
+        # Cleanup
+        await memory.delete_bank(bank_id, request_context=request_context)
+
+    @pytest.mark.asyncio
+    async def test_update_reflection(self, memory: MemoryEngine, request_context):
+        """Test updating a reflection."""
+        bank_id = f"test-reflection-update-{uuid.uuid4().hex[:8]}"
+
+        # Create the bank first
+        await memory.get_bank_profile(bank_id=bank_id, request_context=request_context)
+
+        # Create a reflection
+        reflection = await memory.create_reflection(
+            bank_id=bank_id,
+            name="Original Name",
+            source_query="Original Query",
+            content="Original Content",
+            request_context=request_context,
+        )
+
+        # Update the reflection
+        updated = await memory.update_reflection(
+            bank_id=bank_id,
+            reflection_id=reflection["id"],
+            name="Updated Name",
+            content="Updated Content",
+            request_context=request_context,
+        )
+
+        assert updated["name"] == "Updated Name"
+        assert updated["content"] == "Updated Content"
+
+        # Cleanup
+        await memory.delete_bank(bank_id, request_context=request_context)
+
+    @pytest.mark.asyncio
+    async def test_delete_reflection(self, memory: MemoryEngine, request_context):
+        """Test deleting a reflection."""
+        bank_id = f"test-reflection-delete-{uuid.uuid4().hex[:8]}"
+
+        # Create the bank first
+        await memory.get_bank_profile(bank_id=bank_id, request_context=request_context)
+
+        # Create a reflection
+        reflection = await memory.create_reflection(
+            bank_id=bank_id,
+            name="To Delete",
+            source_query="Query",
+            content="Content",
+            request_context=request_context,
+        )
+
+        # Delete the reflection
+        await memory.delete_reflection(
+            bank_id=bank_id,
+            reflection_id=reflection["id"],
+            request_context=request_context,
+        )
+
+        # Verify deletion - should return None
+        fetched = await memory.get_reflection(
+            bank_id=bank_id,
+            reflection_id=reflection["id"],
+            request_context=request_context,
+        )
+        assert fetched is None
+
+        # Cleanup
+        await memory.delete_bank(bank_id, request_context=request_context)
+
+
+class TestMentalModelsAPI:
+    """Test mental models API endpoints.
+
+    NOTE: Mental models are now stored in memory_units with fact_type='mental_model'
+    and accessed via recall with fact_type=["mental_model"]. The old /mental-models
+    endpoint was removed. These tests are skipped.
+    """
+
+    @pytest.mark.skip(reason="Mental models endpoint removed - use recall with fact_type=['mental_model']")
+    @pytest.mark.asyncio
+    async def test_list_mental_models_empty(self, api_client, test_bank_id):
+        """Test listing mental models when none exist."""
+        pass
+
+    @pytest.mark.skip(reason="Mental models endpoint removed - use recall with fact_type=['mental_model']")
+    @pytest.mark.asyncio
+    async def test_get_mental_model_not_found(self, api_client, test_bank_id):
+        """Test getting a non-existent mental model."""
+        pass
+
+
+class TestReflectionsAPI:
+    """Test reflections API endpoints."""
+
+    @pytest.mark.asyncio
+    async def test_reflections_api_crud(self, api_client, test_bank_id):
+        """Test full CRUD cycle through API."""
+        import asyncio
+
+        # Create bank first via profile endpoint
+        await api_client.get(f"/v1/default/banks/{test_bank_id}/profile")
+
+        # Create a reflection (async operation)
+        response = await api_client.post(
+            f"/v1/default/banks/{test_bank_id}/reflections",
+            json={
+                "name": "API Test Reflection",
+                "source_query": "What is the API test about?",
+                "content": "This is an API test reflection",
+                "tags": ["api-test"],
+            },
+        )
+        assert response.status_code == 200
+        create_result = response.json()
+        assert "operation_id" in create_result
+        operation_id = create_result["operation_id"]
+
+        # Wait for the async operation to complete
+        for _ in range(30):  # Wait up to 30 seconds
+            response = await api_client.get(f"/v1/default/banks/{test_bank_id}/operations/{operation_id}")
+            if response.status_code == 200:
+                op_status = response.json()
+                if op_status.get("status") == "completed":
+                    break
+            await asyncio.sleep(1)
+
+        # List reflections to get the created reflection
+        response = await api_client.get(f"/v1/default/banks/{test_bank_id}/reflections")
+        assert response.status_code == 200
+        reflections = response.json()["items"]
+        assert len(reflections) >= 1
+
+        # Find our reflection
+        reflection = next((r for r in reflections if r["name"] == "API Test Reflection"), None)
+        assert reflection is not None, f"Reflection not found. Items: {reflections}"
+        reflection_id = reflection["id"]
+
+        # Get the reflection
+        response = await api_client.get(f"/v1/default/banks/{test_bank_id}/reflections/{reflection_id}")
+        assert response.status_code == 200
+        assert response.json()["name"] == "API Test Reflection"
+
+        # Update the reflection
+        response = await api_client.patch(
+            f"/v1/default/banks/{test_bank_id}/reflections/{reflection_id}",
+            json={"name": "Updated API Test Reflection"},
+        )
+        assert response.status_code == 200
+        assert response.json()["name"] == "Updated API Test Reflection"
+
+        # Delete the reflection
+        response = await api_client.delete(f"/v1/default/banks/{test_bank_id}/reflections/{reflection_id}")
+        assert response.status_code == 200
+
+        # Verify deletion
+        response = await api_client.get(f"/v1/default/banks/{test_bank_id}/reflections/{reflection_id}")
+        assert response.status_code == 404
+
+        # Cleanup
+        await api_client.delete(f"/v1/default/banks/{test_bank_id}")
+
+
+class TestRecallWithMentalModelsAndReflections:
+    """Test recall integration with mental models and reflections."""
+
+    @pytest.mark.asyncio
+    async def test_recall_includes_mental_models(self, api_client, test_bank_id):
+        """Test that recall can include mental models in the response."""
+        # Create bank first via profile endpoint
+        await api_client.get(f"/v1/default/banks/{test_bank_id}/profile")
+
+        # Note: Mental models are auto-created via consolidation, not manually
+        # This test just verifies the include parameter works
+
+        # Recall with mental models included
+        response = await api_client.post(
+            f"/v1/default/banks/{test_bank_id}/memories/recall",
+            json={
+                "query": "What is machine learning?",
+                "include": {
+                    "mental_models": {"max_results": 5},
+                },
+            },
+        )
+        assert response.status_code == 200
+        result = response.json()
+
+        # Should have mental_models field in response (may be empty)
+        assert "mental_models" in result or result.get("mental_models") is None
+
+        # Cleanup
+        await api_client.delete(f"/v1/default/banks/{test_bank_id}")
+
+    @pytest.mark.asyncio
+    async def test_recall_includes_reflections(self, api_client, test_bank_id):
+        """Test that recall can include reflections in the response."""
+        # Create bank first via profile endpoint
+        await api_client.get(f"/v1/default/banks/{test_bank_id}/profile")
+
+        # Create a reflection first
+        response = await api_client.post(
+            f"/v1/default/banks/{test_bank_id}/reflections",
+            json={
+                "name": "AI Overview",
+                "source_query": "What is AI?",
+                "content": "Artificial intelligence is the simulation of human intelligence",
+                "tags": [],
+            },
+        )
+        assert response.status_code == 200
+
+        # Recall with reflections included
+        response = await api_client.post(
+            f"/v1/default/banks/{test_bank_id}/memories/recall",
+            json={
+                "query": "What is artificial intelligence?",
+                "include": {
+                    "reflections": {"max_results": 5},
+                },
+            },
+        )
+        assert response.status_code == 200
+        result = response.json()
+
+        # Should have reflections in response (may be empty if embedding not generated yet)
+        assert "reflections" in result or result.get("reflections") is None
+
+        # Cleanup
+        await api_client.delete(f"/v1/default/banks/{test_bank_id}")
+
+    @pytest.mark.asyncio
+    async def test_recall_without_mental_models_by_default(self, api_client, test_bank_id):
+        """Test that recall does not include mental models by default."""
+        # Create bank first via profile endpoint
+        await api_client.get(f"/v1/default/banks/{test_bank_id}/profile")
+
+        # Recall without specifying mental models
+        response = await api_client.post(
+            f"/v1/default/banks/{test_bank_id}/memories/recall",
+            json={
+                "query": "Test query",
+            },
+        )
+        assert response.status_code == 200
+        result = response.json()
+
+        # Mental models should not be in response
+        assert result.get("mental_models") is None
+
+        # Cleanup
+        await api_client.delete(f"/v1/default/banks/{test_bank_id}")

hindsight-api/tests/test_server_module.py

@@ -257,7 +257,6 @@ from hindsight_api.extensions import (
     RetainContext,
     RecallContext,
     ReflectContext,
-    RefreshMentalModelContext,
 )
 
 
@@ -289,6 +288,3 @@ class MockOperationValidator(OperationValidatorExtension):
 
     async def validate_reflect(self, ctx: ReflectContext) -> ValidationResult:
         return ValidationResult.accept()
-
-    async def validate_refresh_mental_model(self, ctx: RefreshMentalModelContext) -> ValidationResult:
-        return ValidationResult.accept()

hindsight-api/tests/test_sql_schema_safety.py

@@ -21,6 +21,8 @@ TABLES = [
     "documents",
     "chunks",
     "async_operations",
+    "directives",
+    "reflections",
 ]
 
 # Files to scan for SQL queries

hindsight-api/tests/test_worker.py

@@ -0,0 +1,827 @@
+"""
+Tests for the distributed worker system.
+
+Tests cover:
+- BrokerTaskBackend task submission and storage
+- WorkerPoller task claiming with FOR UPDATE SKIP LOCKED
+- Concurrent workers claiming different tasks (no duplicates)
+- Task completion and failure handling
+- Retry mechanism
+- Worker decommissioning
+"""
+
+import asyncio
+import json
+import uuid
+
+import pytest
+import pytest_asyncio
+
+from hindsight_api.engine.task_backend import BrokerTaskBackend, SyncTaskBackend
+
+
+# Use loadgroup to ensure these tests run in the same worker
+# since they share database state
+pytestmark = pytest.mark.xdist_group("worker_tests")
+
+
+@pytest_asyncio.fixture
+async def pool(pg0_db_url):
+    """Create a dedicated connection pool for worker tests."""
+    import asyncpg
+
+    from hindsight_api.pg0 import resolve_database_url
+
+    # Resolve pg0:// URL to postgresql:// URL if needed
+    resolved_url = await resolve_database_url(pg0_db_url)
+
+    pool = await asyncpg.create_pool(
+        resolved_url,
+        min_size=2,
+        max_size=10,
+        command_timeout=30,
+    )
+    yield pool
+    await pool.close()
+
+
+@pytest_asyncio.fixture
+async def clean_operations(pool):
+    """Clean up async_operations table before and after tests."""
+    # Clean before test
+    await pool.execute("DELETE FROM async_operations WHERE bank_id LIKE 'test-worker-%'")
+    yield
+    # Clean after test
+    await pool.execute("DELETE FROM async_operations WHERE bank_id LIKE 'test-worker-%'")
+
+
+class TestBrokerTaskBackend:
+    """Tests for BrokerTaskBackend task storage."""
+
+    @pytest.mark.asyncio
+    async def test_submit_task_updates_existing_operation(self, pool, clean_operations):
+        """Test that submit_task updates task_payload for existing operations."""
+        # Create an operation record first
+        operation_id = uuid.uuid4()
+        bank_id = f"test-worker-{uuid.uuid4().hex[:8]}"
+
+        await pool.execute(
+            """
+            INSERT INTO async_operations (operation_id, bank_id, operation_type, status)
+            VALUES ($1, $2, 'test_operation', 'pending')
+            """,
+            operation_id,
+            bank_id,
+        )
+
+        # Submit task with same operation_id
+        backend = BrokerTaskBackend(pool_getter=lambda: pool)
+        await backend.initialize()
+
+        task_dict = {
+            "operation_id": str(operation_id),
+            "type": "test_task",
+            "bank_id": bank_id,
+            "data": {"key": "value"},
+        }
+        await backend.submit_task(task_dict)
+
+        # Verify task_payload was stored
+        row = await pool.fetchrow(
+            "SELECT task_payload, status FROM async_operations WHERE operation_id = $1",
+            operation_id,
+        )
+        assert row is not None
+        assert row["status"] == "pending"
+        payload = json.loads(row["task_payload"])
+        assert payload["type"] == "test_task"
+        assert payload["data"] == {"key": "value"}
+
+    @pytest.mark.asyncio
+    async def test_submit_task_creates_new_operation(self, pool, clean_operations):
+        """Test that submit_task creates new operation when no operation_id provided."""
+        backend = BrokerTaskBackend(pool_getter=lambda: pool)
+        await backend.initialize()
+
+        bank_id = f"test-worker-{uuid.uuid4().hex[:8]}"
+        task_dict = {
+            "type": "access_count_update",
+            "bank_id": bank_id,
+            "node_ids": ["node1", "node2"],
+        }
+        await backend.submit_task(task_dict)
+
+        # Verify new operation was created
+        row = await pool.fetchrow(
+            "SELECT operation_type, status, task_payload FROM async_operations WHERE bank_id = $1",
+            bank_id,
+        )
+        assert row is not None
+        assert row["operation_type"] == "access_count_update"
+        assert row["status"] == "pending"
+        payload = json.loads(row["task_payload"])
+        assert payload["node_ids"] == ["node1", "node2"]
+
+
+class TestWorkerPoller:
+    """Tests for WorkerPoller task claiming and execution."""
+
+    @pytest.mark.asyncio
+    async def test_claim_batch_claims_pending_tasks(self, pool, clean_operations):
+        """Test that claim_batch claims pending tasks with task_payload."""
+        from hindsight_api.worker import WorkerPoller
+
+        # Create some pending tasks
+        bank_id = f"test-worker-{uuid.uuid4().hex[:8]}"
+        for i in range(3):
+            op_id = uuid.uuid4()
+            payload = json.dumps({"type": "test_task", "index": i, "bank_id": bank_id})
+            await pool.execute(
+                """
+                INSERT INTO async_operations (operation_id, bank_id, operation_type, status, task_payload)
+                VALUES ($1, $2, 'test', 'pending', $3::jsonb)
+                """,
+                op_id,
+                bank_id,
+                payload,
+            )
+
+        # Create poller and claim tasks
+        executed_tasks = []
+
+        async def mock_executor(task_dict):
+            executed_tasks.append(task_dict)
+
+        poller = WorkerPoller(
+            pool=pool,
+            worker_id="test-worker-1",
+            executor=mock_executor,
+            batch_size=10,
+        )
+
+        claimed = await poller.claim_batch()
+        assert len(claimed) == 3
+
+        # Verify tasks are marked as processing with worker_id
+        rows = await pool.fetch(
+            "SELECT status, worker_id FROM async_operations WHERE bank_id = $1",
+            bank_id,
+        )
+        for row in rows:
+            assert row["status"] == "processing"
+            assert row["worker_id"] == "test-worker-1"
+
+    @pytest.mark.asyncio
+    async def test_claim_batch_respects_batch_size(self, pool, clean_operations):
+        """Test that claim_batch respects the batch_size limit."""
+        from hindsight_api.worker import WorkerPoller
+
+        # Create 10 pending tasks
+        bank_id = f"test-worker-{uuid.uuid4().hex[:8]}"
+        for i in range(10):
+            op_id = uuid.uuid4()
+            payload = json.dumps({"type": "test_task", "index": i, "bank_id": bank_id})
+            await pool.execute(
+                """
+                INSERT INTO async_operations (operation_id, bank_id, operation_type, status, task_payload)
+                VALUES ($1, $2, 'test', 'pending', $3::jsonb)
+                """,
+                op_id,
+                bank_id,
+                payload,
+            )
+
+        # Claim with batch_size=3
+        poller = WorkerPoller(
+            pool=pool,
+            worker_id="test-worker-1",
+            executor=lambda x: None,
+            batch_size=3,
+        )
+
+        claimed = await poller.claim_batch()
+        assert len(claimed) == 3
+
+    @pytest.mark.asyncio
+    async def test_execute_task_marks_completed(self, pool, clean_operations):
+        """Test that successful task execution marks task as completed."""
+        from hindsight_api.worker import WorkerPoller
+
+        # Create a pending task
+        bank_id = f"test-worker-{uuid.uuid4().hex[:8]}"
+        op_id = uuid.uuid4()
+        payload = json.dumps({"type": "test_task", "operation_id": str(op_id), "bank_id": bank_id})
+        await pool.execute(
+            """
+            INSERT INTO async_operations (operation_id, bank_id, operation_type, status, task_payload, worker_id)
+            VALUES ($1, $2, 'test', 'processing', $3::jsonb, 'test-worker-1')
+            """,
+            op_id,
+            bank_id,
+            payload,
+        )
+
+        executed = []
+
+        async def mock_executor(task_dict):
+            executed.append(task_dict)
+
+        poller = WorkerPoller(
+            pool=pool,
+            worker_id="test-worker-1",
+            executor=mock_executor,
+        )
+
+        # Execute the task
+        task_dict = json.loads(payload)
+        await poller.execute_task(str(op_id), task_dict)
+
+        assert len(executed) == 1
+
+        # Verify task is marked as completed
+        row = await pool.fetchrow(
+            "SELECT status, completed_at FROM async_operations WHERE operation_id = $1",
+            op_id,
+        )
+        assert row["status"] == "completed"
+        assert row["completed_at"] is not None
+
+    @pytest.mark.asyncio
+    async def test_execute_task_retries_on_failure(self, pool, clean_operations):
+        """Test that failed task execution triggers retry mechanism."""
+        from hindsight_api.worker import WorkerPoller
+
+        # Create a pending task with retry_count=0
+        bank_id = f"test-worker-{uuid.uuid4().hex[:8]}"
+        op_id = uuid.uuid4()
+        payload = json.dumps({"type": "test_task", "operation_id": str(op_id), "bank_id": bank_id})
+        await pool.execute(
+            """
+            INSERT INTO async_operations (operation_id, bank_id, operation_type, status, task_payload, worker_id, retry_count)
+            VALUES ($1, $2, 'test', 'processing', $3::jsonb, 'test-worker-1', 0)
+            """,
+            op_id,
+            bank_id,
+            payload,
+        )
+
+        async def failing_executor(task_dict):
+            raise ValueError("Simulated failure")
+
+        poller = WorkerPoller(
+            pool=pool,
+            worker_id="test-worker-1",
+            executor=failing_executor,
+            max_retries=3,
+        )
+
+        # Execute (should fail and retry)
+        task_dict = json.loads(payload)
+        await poller.execute_task(str(op_id), task_dict)
+
+        # Verify task is back to pending with incremented retry_count
+        row = await pool.fetchrow(
+            "SELECT status, retry_count, worker_id FROM async_operations WHERE operation_id = $1",
+            op_id,
+        )
+        assert row["status"] == "pending"
+        assert row["retry_count"] == 1
+        assert row["worker_id"] is None  # Worker ID cleared for retry
+
+    @pytest.mark.asyncio
+    async def test_execute_task_fails_after_max_retries(self, pool, clean_operations):
+        """Test that task is marked failed after exceeding max retries."""
+        from hindsight_api.worker import WorkerPoller
+
+        # Create a task that has already used all retries
+        bank_id = f"test-worker-{uuid.uuid4().hex[:8]}"
+        op_id = uuid.uuid4()
+        payload = json.dumps({"type": "test_task", "operation_id": str(op_id), "bank_id": bank_id})
+        await pool.execute(
+            """
+            INSERT INTO async_operations (operation_id, bank_id, operation_type, status, task_payload, worker_id, retry_count)
+            VALUES ($1, $2, 'test', 'processing', $3::jsonb, 'test-worker-1', 3)
+            """,
+            op_id,
+            bank_id,
+            payload,
+        )
+
+        async def failing_executor(task_dict):
+            raise ValueError("Simulated failure")
+
+        poller = WorkerPoller(
+            pool=pool,
+            worker_id="test-worker-1",
+            executor=failing_executor,
+            max_retries=3,
+        )
+
+        # Execute (should fail permanently)
+        task_dict = json.loads(payload)
+        await poller.execute_task(str(op_id), task_dict)
+
+        # Verify task is marked as failed
+        row = await pool.fetchrow(
+            "SELECT status, error_message FROM async_operations WHERE operation_id = $1",
+            op_id,
+        )
+        assert row["status"] == "failed"
+        assert "Max retries" in row["error_message"]
+
+    @pytest.mark.asyncio
+    async def test_claim_batch_skips_consolidation_when_same_bank_processing(self, pool, clean_operations):
+        """Test that pending consolidation is skipped if same bank has one processing."""
+        from hindsight_api.worker import WorkerPoller
+
+        bank_id = f"test-worker-{uuid.uuid4().hex[:8]}"
+
+        # Create a processing consolidation for bank
+        processing_op_id = uuid.uuid4()
+        await pool.execute(
+            """
+            INSERT INTO async_operations (operation_id, bank_id, operation_type, status, task_payload, worker_id)
+            VALUES ($1, $2, 'consolidation', 'processing', $3::jsonb, 'other-worker')
+            """,
+            processing_op_id,
+            bank_id,
+            json.dumps({"type": "consolidation", "bank_id": bank_id}),
+        )
+
+        # Create a pending consolidation for same bank (should be skipped)
+        pending_op_id = uuid.uuid4()
+        await pool.execute(
+            """
+            INSERT INTO async_operations (operation_id, bank_id, operation_type, status, task_payload)
+            VALUES ($1, $2, 'consolidation', 'pending', $3::jsonb)
+            """,
+            pending_op_id,
+            bank_id,
+            json.dumps({"type": "consolidation", "bank_id": bank_id}),
+        )
+
+        # Create a pending consolidation for different bank (should be claimed)
+        other_bank_id = f"test-worker-{uuid.uuid4().hex[:8]}"
+        other_op_id = uuid.uuid4()
+        await pool.execute(
+            """
+            INSERT INTO async_operations (operation_id, bank_id, operation_type, status, task_payload)
+            VALUES ($1, $2, 'consolidation', 'pending', $3::jsonb)
+            """,
+            other_op_id,
+            other_bank_id,
+            json.dumps({"type": "consolidation", "bank_id": other_bank_id}),
+        )
+
+        poller = WorkerPoller(
+            pool=pool,
+            worker_id="test-worker-1",
+            executor=lambda x: None,
+            batch_size=10,
+        )
+
+        claimed = await poller.claim_batch()
+
+        # Should only claim the consolidation for the other bank
+        assert len(claimed) == 1
+        claimed_op_id, claimed_payload = claimed[0]
+        assert claimed_op_id == str(other_op_id)
+        assert claimed_payload["bank_id"] == other_bank_id
+
+        # Verify the pending consolidation for first bank is still pending
+        row = await pool.fetchrow(
+            "SELECT status, worker_id FROM async_operations WHERE operation_id = $1",
+            pending_op_id,
+        )
+        assert row["status"] == "pending"
+        assert row["worker_id"] is None
+
+    @pytest.mark.asyncio
+    async def test_claim_batch_allows_non_consolidation_when_consolidation_processing(self, pool, clean_operations):
+        """Test that non-consolidation tasks are still claimed even if consolidation is processing."""
+        from hindsight_api.worker import WorkerPoller
+
+        bank_id = f"test-worker-{uuid.uuid4().hex[:8]}"
+
+        # Create a processing consolidation for bank
+        await pool.execute(
+            """
+            INSERT INTO async_operations (operation_id, bank_id, operation_type, status, task_payload, worker_id)
+            VALUES ($1, $2, 'consolidation', 'processing', $3::jsonb, 'other-worker')
+            """,
+            uuid.uuid4(),
+            bank_id,
+            json.dumps({"type": "consolidation", "bank_id": bank_id}),
+        )
+
+        # Create a pending retain task for same bank (should be claimed)
+        retain_op_id = uuid.uuid4()
+        await pool.execute(
+            """
+            INSERT INTO async_operations (operation_id, bank_id, operation_type, status, task_payload)
+            VALUES ($1, $2, 'retain', 'pending', $3::jsonb)
+            """,
+            retain_op_id,
+            bank_id,
+            json.dumps({"type": "batch_retain", "bank_id": bank_id}),
+        )
+
+        poller = WorkerPoller(
+            pool=pool,
+            worker_id="test-worker-1",
+            executor=lambda x: None,
+            batch_size=10,
+        )
+
+        claimed = await poller.claim_batch()
+
+        # Should claim the retain task (non-consolidation tasks are unaffected)
+        assert len(claimed) == 1
+        claimed_op_id, _ = claimed[0]
+        assert claimed_op_id == str(retain_op_id)
+
+
+class TestWorkerRecovery:
+    """Tests for worker task recovery on startup."""
+
+    @pytest.mark.asyncio
+    async def test_recover_own_tasks_resets_processing_to_pending(self, pool, clean_operations):
+        """Test that recover_own_tasks resets processing tasks back to pending."""
+        from hindsight_api.worker import WorkerPoller
+
+        # Create tasks that were being processed by this worker (simulating a crash)
+        bank_id = f"test-worker-{uuid.uuid4().hex[:8]}"
+        worker_id = "crashed-worker"
+        task_ids = []
+
+        for i in range(3):
+            op_id = uuid.uuid4()
+            task_ids.append(op_id)
+            payload = json.dumps({"type": "test_task", "index": i, "bank_id": bank_id})
+            await pool.execute(
+                """
+                INSERT INTO async_operations (operation_id, bank_id, operation_type, status, task_payload, worker_id, claimed_at)
+                VALUES ($1, $2, 'test', 'processing', $3::jsonb, $4, now())
+                """,
+                op_id,
+                bank_id,
+                payload,
+                worker_id,
+            )
+
+        # Create poller with same worker_id and call recover
+        poller = WorkerPoller(
+            pool=pool,
+            worker_id=worker_id,
+            executor=lambda x: None,
+        )
+
+        recovered_count = await poller.recover_own_tasks()
+        assert recovered_count == 3
+
+        # Verify all tasks are back to pending with no worker assigned
+        rows = await pool.fetch(
+            "SELECT status, worker_id, claimed_at FROM async_operations WHERE bank_id = $1",
+            bank_id,
+        )
+        for row in rows:
+            assert row["status"] == "pending"
+            assert row["worker_id"] is None
+            assert row["claimed_at"] is None
+
+    @pytest.mark.asyncio
+    async def test_recover_own_tasks_does_not_affect_other_workers(self, pool, clean_operations):
+        """Test that recover_own_tasks only affects tasks from the same worker_id."""
+        from hindsight_api.worker import WorkerPoller
+
+        bank_id = f"test-worker-{uuid.uuid4().hex[:8]}"
+
+        # Create tasks for worker-1 (the one that will recover)
+        for i in range(2):
+            op_id = uuid.uuid4()
+            payload = json.dumps({"type": "test_task", "index": i, "bank_id": bank_id})
+            await pool.execute(
+                """
+                INSERT INTO async_operations (operation_id, bank_id, operation_type, status, task_payload, worker_id)
+                VALUES ($1, $2, 'test', 'processing', $3::jsonb, 'worker-1')
+                """,
+                op_id,
+                bank_id,
+                payload,
+            )
+
+        # Create tasks for worker-2 (should not be affected)
+        for i in range(2):
+            op_id = uuid.uuid4()
+            payload = json.dumps({"type": "test_task", "index": i + 10, "bank_id": bank_id})
+            await pool.execute(
+                """
+                INSERT INTO async_operations (operation_id, bank_id, operation_type, status, task_payload, worker_id)
+                VALUES ($1, $2, 'test', 'processing', $3::jsonb, 'worker-2')
+                """,
+                op_id,
+                bank_id,
+                payload,
+            )
+
+        # Worker-1 recovers its tasks
+        poller = WorkerPoller(
+            pool=pool,
+            worker_id="worker-1",
+            executor=lambda x: None,
+        )
+
+        recovered_count = await poller.recover_own_tasks()
+        assert recovered_count == 2
+
+        # Verify worker-1 tasks are released
+        worker1_rows = await pool.fetch(
+            "SELECT status, worker_id FROM async_operations WHERE bank_id = $1 AND worker_id IS NULL",
+            bank_id,
+        )
+        assert len(worker1_rows) == 2
+
+        # Verify worker-2 tasks are unaffected
+        worker2_rows = await pool.fetch(
+            "SELECT status, worker_id FROM async_operations WHERE bank_id = $1 AND worker_id = 'worker-2'",
+            bank_id,
+        )
+        assert len(worker2_rows) == 2
+        for row in worker2_rows:
+            assert row["status"] == "processing"
+
+    @pytest.mark.asyncio
+    async def test_recover_own_tasks_returns_zero_when_no_stale_tasks(self, pool, clean_operations):
+        """Test that recover_own_tasks returns 0 when there are no stale tasks."""
+        from hindsight_api.worker import WorkerPoller
+
+        poller = WorkerPoller(
+            pool=pool,
+            worker_id="fresh-worker",
+            executor=lambda x: None,
+        )
+
+        recovered_count = await poller.recover_own_tasks()
+        assert recovered_count == 0
+
+
+class TestConcurrentWorkers:
+    """Tests for concurrent worker task claiming (FOR UPDATE SKIP LOCKED)."""
+
+    @pytest.mark.asyncio
+    async def test_concurrent_workers_claim_different_tasks(self, pool, clean_operations):
+        """Test that multiple workers claim different tasks (no duplicates)."""
+        from hindsight_api.worker import WorkerPoller
+
+        # Create 10 pending tasks
+        bank_id = f"test-worker-{uuid.uuid4().hex[:8]}"
+        task_ids = []
+        for i in range(10):
+            op_id = uuid.uuid4()
+            task_ids.append(op_id)
+            payload = json.dumps({"type": "test_task", "index": i, "bank_id": bank_id, "operation_id": str(op_id)})
+            await pool.execute(
+                """
+                INSERT INTO async_operations (operation_id, bank_id, operation_type, status, task_payload)
+                VALUES ($1, $2, 'test', 'pending', $3::jsonb)
+                """,
+                op_id,
+                bank_id,
+                payload,
+            )
+
+        # Create 3 workers that will claim tasks concurrently
+        workers_claimed: dict[str, list[str]] = {"worker-1": [], "worker-2": [], "worker-3": []}
+
+        async def claim_for_worker(worker_id: str):
+            poller = WorkerPoller(
+                pool=pool,
+                worker_id=worker_id,
+                executor=lambda x: None,
+                batch_size=5,  # Each worker tries to claim 5
+            )
+            claimed = await poller.claim_batch()
+            workers_claimed[worker_id] = [op_id for op_id, _ in claimed]
+
+        # Run all workers concurrently
+        await asyncio.gather(
+            claim_for_worker("worker-1"),
+            claim_for_worker("worker-2"),
+            claim_for_worker("worker-3"),
+        )
+
+        # Verify no duplicates - each task claimed by exactly one worker
+        all_claimed = workers_claimed["worker-1"] + workers_claimed["worker-2"] + workers_claimed["worker-3"]
+        assert len(all_claimed) == len(set(all_claimed)), "Duplicate task claimed by multiple workers!"
+
+        # Verify total claimed equals available tasks (10)
+        assert len(all_claimed) == 10, f"Expected 10 tasks claimed, got {len(all_claimed)}"
+
+        # Verify each task is assigned to exactly one worker in DB
+        rows = await pool.fetch(
+            "SELECT operation_id, worker_id FROM async_operations WHERE bank_id = $1",
+            bank_id,
+        )
+        worker_assignments = {str(row["operation_id"]): row["worker_id"] for row in rows}
+
+        # With FOR UPDATE SKIP LOCKED, it's a race condition which workers get tasks.
+        # The important invariant is no duplicates and all tasks claimed, which we verified above.
+        # Just verify that at least 1 worker got tasks and all tasks have a worker assigned.
+        assert len(set(worker_assignments.values())) >= 1, "At least one worker should have claimed tasks"
+        assert all(w is not None for w in worker_assignments.values()), "All tasks should have a worker assigned"
+
+    @pytest.mark.asyncio
+    async def test_workers_do_not_claim_already_processing_tasks(self, pool, clean_operations):
+        """Test that workers skip tasks already being processed by another worker."""
+        from hindsight_api.worker import WorkerPoller
+
+        # Create tasks - some pending, some already processing
+        bank_id = f"test-worker-{uuid.uuid4().hex[:8]}"
+
+        # Create 3 pending tasks
+        for i in range(3):
+            op_id = uuid.uuid4()
+            payload = json.dumps({"type": "test_task", "index": i, "bank_id": bank_id})
+            await pool.execute(
+                """
+                INSERT INTO async_operations (operation_id, bank_id, operation_type, status, task_payload)
+                VALUES ($1, $2, 'test', 'pending', $3::jsonb)
+                """,
+                op_id,
+                bank_id,
+                payload,
+            )
+
+        # Create 2 already-processing tasks owned by another worker
+        for i in range(2):
+            op_id = uuid.uuid4()
+            payload = json.dumps({"type": "test_task", "index": i + 10, "bank_id": bank_id})
+            await pool.execute(
+                """
+                INSERT INTO async_operations (operation_id, bank_id, operation_type, status, task_payload, worker_id)
+                VALUES ($1, $2, 'test', 'processing', $3::jsonb, 'other-worker')
+                """,
+                op_id,
+                bank_id,
+                payload,
+            )
+
+        # New worker should only claim the 3 pending tasks
+        poller = WorkerPoller(
+            pool=pool,
+            worker_id="new-worker",
+            executor=lambda x: None,
+            batch_size=10,
+        )
+
+        claimed = await poller.claim_batch()
+        assert len(claimed) == 3, "Worker should only claim pending tasks"
+
+        # Verify other worker's tasks are still owned by them
+        row = await pool.fetchrow(
+            "SELECT COUNT(*) as count FROM async_operations WHERE bank_id = $1 AND worker_id = 'other-worker'",
+            bank_id,
+        )
+        assert row["count"] == 2
+
+
+class TestWorkerDecommission:
+    """Tests for worker decommissioning functionality."""
+
+    @pytest.mark.asyncio
+    async def test_decommission_releases_worker_tasks(self, pool, clean_operations):
+        """Test that decommissioning a worker releases all its processing tasks."""
+        # Create tasks being processed by a worker
+        bank_id = f"test-worker-{uuid.uuid4().hex[:8]}"
+        worker_id = "worker-to-decommission"
+
+        for i in range(5):
+            op_id = uuid.uuid4()
+            payload = json.dumps({"type": "test_task", "index": i, "bank_id": bank_id})
+            await pool.execute(
+                """
+                INSERT INTO async_operations (operation_id, bank_id, operation_type, status, task_payload, worker_id, claimed_at)
+                VALUES ($1, $2, 'test', 'processing', $3::jsonb, $4, now())
+                """,
+                op_id,
+                bank_id,
+                payload,
+                worker_id,
+            )
+
+        # Run decommission
+        result = await pool.fetch(
+            """
+            UPDATE async_operations
+            SET status = 'pending', worker_id = NULL, claimed_at = NULL, updated_at = now()
+            WHERE worker_id = $1 AND status = 'processing'
+            RETURNING operation_id
+            """,
+            worker_id,
+        )
+
+        assert len(result) == 5
+
+        # Verify all tasks are back to pending
+        rows = await pool.fetch(
+            "SELECT status, worker_id, claimed_at FROM async_operations WHERE bank_id = $1",
+            bank_id,
+        )
+        for row in rows:
+            assert row["status"] == "pending"
+            assert row["worker_id"] is None
+            assert row["claimed_at"] is None
+
+    @pytest.mark.asyncio
+    async def test_decommission_does_not_affect_other_workers(self, pool, clean_operations):
+        """Test that decommissioning one worker doesn't affect another worker's tasks."""
+        bank_id = f"test-worker-{uuid.uuid4().hex[:8]}"
+
+        # Create tasks for worker-1
+        for i in range(3):
+            op_id = uuid.uuid4()
+            payload = json.dumps({"type": "test_task", "index": i, "bank_id": bank_id})
+            await pool.execute(
+                """
+                INSERT INTO async_operations (operation_id, bank_id, operation_type, status, task_payload, worker_id)
+                VALUES ($1, $2, 'test', 'processing', $3::jsonb, 'worker-1')
+                """,
+                op_id,
+                bank_id,
+                payload,
+            )
+
+        # Create tasks for worker-2
+        for i in range(3):
+            op_id = uuid.uuid4()
+            payload = json.dumps({"type": "test_task", "index": i + 10, "bank_id": bank_id})
+            await pool.execute(
+                """
+                INSERT INTO async_operations (operation_id, bank_id, operation_type, status, task_payload, worker_id)
+                VALUES ($1, $2, 'test', 'processing', $3::jsonb, 'worker-2')
+                """,
+                op_id,
+                bank_id,
+                payload,
+            )
+
+        # Decommission worker-1 only
+        await pool.execute(
+            """
+            UPDATE async_operations
+            SET status = 'pending', worker_id = NULL, claimed_at = NULL
+            WHERE worker_id = 'worker-1' AND status = 'processing'
+            """
+        )
+
+        # Verify worker-1 tasks are released
+        worker1_rows = await pool.fetch(
+            "SELECT status, worker_id FROM async_operations WHERE bank_id = $1 AND worker_id IS NULL",
+            bank_id,
+        )
+        assert len(worker1_rows) == 3
+
+        # Verify worker-2 tasks are unaffected
+        worker2_rows = await pool.fetch(
+            "SELECT status, worker_id FROM async_operations WHERE bank_id = $1 AND worker_id = 'worker-2'",
+            bank_id,
+        )
+        assert len(worker2_rows) == 3
+        for row in worker2_rows:
+            assert row["status"] == "processing"
+
+
+class TestSyncTaskBackend:
+    """Tests for SyncTaskBackend (used in tests and embedded mode)."""
+
+    @pytest.mark.asyncio
+    async def test_sync_backend_executes_immediately(self):
+        """Test that SyncTaskBackend executes tasks immediately."""
+        executed = []
+
+        async def mock_executor(task_dict):
+            executed.append(task_dict)
+
+        backend = SyncTaskBackend()
+        backend.set_executor(mock_executor)
+        await backend.initialize()
+
+        task_dict = {"type": "test", "data": "value"}
+        await backend.submit_task(task_dict)
+
+        assert len(executed) == 1
+        assert executed[0] == task_dict
+
+    @pytest.mark.asyncio
+    async def test_sync_backend_handles_errors(self):
+        """Test that SyncTaskBackend handles executor errors gracefully."""
+
+        async def failing_executor(task_dict):
+            raise ValueError("Test error")
+
+        backend = SyncTaskBackend()
+        backend.set_executor(failing_executor)
+        await backend.initialize()
+
+        # Should not raise, error is logged
+        await backend.submit_task({"type": "test"})

hindsight-cli/Cargo.toml

@@ -45,6 +45,10 @@ chrono = "0.4"
 walkdir = "2.5"
 dirs = "5.0"
 
+[dev-dependencies]
+# For integration tests with blocking HTTP client
+reqwest = { version = "0.12", features = ["blocking"] }
+
 [profile.release]
 opt-level = "z"
 lto = true

hindsight-cli/smoke-test.sh

@@ -67,7 +67,7 @@ run_test_output() {
 cleanup() {
     echo ""
     echo "Cleaning up test bank..."
-    "$HINDSIGHT_CLI" bank delete "$TEST_BANK" 2>/dev/null || true
+    "$HINDSIGHT_CLI" bank delete "$TEST_BANK" -y 2>/dev/null || true
 }
 trap cleanup EXIT
 
@@ -115,8 +115,11 @@ run_test "list documents" "$HINDSIGHT_CLI" document list "$TEST_BANK" || FAILED=
 # Test 14: Clear memories
 run_test "clear memories" "$HINDSIGHT_CLI" memory clear "$TEST_BANK" || FAILED=1
 
-# Test 15: Delete bank
-run_test "delete bank" "$HINDSIGHT_CLI" bank delete "$TEST_BANK" || FAILED=1
+# Test 15: List operations
+run_test "list operations" "$HINDSIGHT_CLI" operation list "$TEST_BANK" || FAILED=1
+
+# Test 16: Delete bank
+run_test "delete bank" "$HINDSIGHT_CLI" bank delete "$TEST_BANK" -y || FAILED=1
 
 echo ""
 if [ $FAILED -eq 0 ]; then

hindsight-cli/src/api.rs

@@ -173,7 +173,7 @@ impl ApiClient {
     pub fn poll_operation(&self, agent_id: &str, operation_id: &str, verbose: bool) -> Result<(bool, Option<String>)> {
         self.runtime.block_on(async {
             loop {
-                let response = self.client.list_operations(agent_id, None).await?;
+                let response = self.client.list_operations(agent_id, None, None, None, None).await?;
                 let ops = response.into_inner();
 
                 // Find our operation
@@ -258,7 +258,7 @@ impl ApiClient {
 
     pub fn list_operations(&self, agent_id: &str, _verbose: bool) -> Result<OperationsResponse> {
         self.runtime.block_on(async {
-            let response = self.client.list_operations(agent_id, None).await?;
+            let response = self.client.list_operations(agent_id, None, None, None, None).await?;
             let value = response.into_inner();
             // Convert to JSON Value first, then parse into our type
             let json_value = serde_json::to_value(&value)?;
@@ -316,6 +316,231 @@ impl ApiClient {
     }
 }
 
+// ============================================================================
+// Additional API methods for complete CLI coverage
+// ============================================================================
+
+impl ApiClient {
+    // --- Memory Methods ---
+
+    pub fn get_memory(&self, bank_id: &str, memory_id: &str, _verbose: bool) -> Result<serde_json::Value> {
+        self.runtime.block_on(async {
+            let response = self.client.get_memory(bank_id, memory_id, None).await?;
+            Ok(response.into_inner())
+        })
+    }
+
+    // --- Bank Methods ---
+
+    pub fn create_bank(
+        &self,
+        bank_id: &str,
+        request: &types::CreateBankRequest,
+        _verbose: bool,
+    ) -> Result<types::BankProfileResponse> {
+        self.runtime.block_on(async {
+            let response = self.client.create_or_update_bank(bank_id, None, request).await?;
+            Ok(response.into_inner())
+        })
+    }
+
+    pub fn update_bank(
+        &self,
+        bank_id: &str,
+        request: &types::CreateBankRequest,
+        _verbose: bool,
+    ) -> Result<types::BankProfileResponse> {
+        self.runtime.block_on(async {
+            let response = self.client.update_bank(bank_id, None, request).await?;
+            Ok(response.into_inner())
+        })
+    }
+
+    pub fn set_mission(
+        &self,
+        bank_id: &str,
+        mission: &str,
+        _verbose: bool,
+    ) -> Result<types::BankProfileResponse> {
+        self.runtime.block_on(async {
+            let request = types::CreateBankRequest {
+                name: None,
+                mission: Some(mission.to_string()),
+                background: None,
+                disposition: None,
+            };
+            let response = self.client.update_bank(bank_id, None, &request).await?;
+            Ok(response.into_inner())
+        })
+    }
+
+    pub fn get_graph(
+        &self,
+        bank_id: &str,
+        type_filter: Option<&str>,
+        limit: Option<i64>,
+        _verbose: bool,
+    ) -> Result<types::GraphDataResponse> {
+        self.runtime.block_on(async {
+            let response = self.client.get_graph(bank_id, limit, type_filter, None).await?;
+            Ok(response.into_inner())
+        })
+    }
+
+    // --- Tag Methods ---
+
+    pub fn list_tags(
+        &self,
+        bank_id: &str,
+        q: Option<&str>,
+        limit: Option<i64>,
+        offset: Option<i64>,
+        _verbose: bool,
+    ) -> Result<types::ListTagsResponse> {
+        self.runtime.block_on(async {
+            let response = self.client.list_tags(bank_id, limit, offset, q, None).await?;
+            Ok(response.into_inner())
+        })
+    }
+
+    // --- Chunk Methods ---
+
+    pub fn get_chunk(&self, chunk_id: &str, _verbose: bool) -> Result<types::ChunkResponse> {
+        self.runtime.block_on(async {
+            let response = self.client.get_chunk(chunk_id, None).await?;
+            Ok(response.into_inner())
+        })
+    }
+
+    // --- Operation Methods ---
+
+    pub fn get_operation(&self, bank_id: &str, operation_id: &str, _verbose: bool) -> Result<types::OperationStatusResponse> {
+        self.runtime.block_on(async {
+            let response = self.client.get_operation_status(bank_id, operation_id, None).await?;
+            Ok(response.into_inner())
+        })
+    }
+
+    // --- Health Methods ---
+
+    pub fn health(&self, _verbose: bool) -> Result<serde_json::Value> {
+        self.runtime.block_on(async {
+            let response = self.client.health_endpoint_health_get().await?;
+            Ok(response.into_inner())
+        })
+    }
+
+    pub fn metrics(&self, _verbose: bool) -> Result<serde_json::Value> {
+        self.runtime.block_on(async {
+            let response = self.client.metrics_endpoint_metrics_get().await?;
+            Ok(response.into_inner())
+        })
+    }
+
+    // --- Reflection Methods ---
+
+    pub fn list_reflections(&self, bank_id: &str, _verbose: bool) -> Result<types::ReflectionListResponse> {
+        self.runtime.block_on(async {
+            let response = self.client.list_reflections(bank_id, None, None, None, None, None).await?;
+            Ok(response.into_inner())
+        })
+    }
+
+    pub fn get_reflection(&self, bank_id: &str, reflection_id: &str, _verbose: bool) -> Result<types::ReflectionResponse> {
+        self.runtime.block_on(async {
+            let response = self.client.get_reflection(bank_id, reflection_id, None).await?;
+            Ok(response.into_inner())
+        })
+    }
+
+    pub fn create_reflection(
+        &self,
+        bank_id: &str,
+        request: &types::CreateReflectionRequest,
+        _verbose: bool,
+    ) -> Result<types::CreateReflectionResponse> {
+        self.runtime.block_on(async {
+            let response = self.client.create_reflection(bank_id, None, request).await?;
+            Ok(response.into_inner())
+        })
+    }
+
+    pub fn update_reflection(
+        &self,
+        bank_id: &str,
+        reflection_id: &str,
+        request: &types::UpdateReflectionRequest,
+        _verbose: bool,
+    ) -> Result<types::ReflectionResponse> {
+        self.runtime.block_on(async {
+            let response = self.client.update_reflection(bank_id, reflection_id, None, request).await?;
+            Ok(response.into_inner())
+        })
+    }
+
+    pub fn delete_reflection(&self, bank_id: &str, reflection_id: &str, _verbose: bool) -> Result<serde_json::Value> {
+        self.runtime.block_on(async {
+            let response = self.client.delete_reflection(bank_id, reflection_id, None).await?;
+            Ok(response.into_inner())
+        })
+    }
+
+    pub fn refresh_reflection(&self, bank_id: &str, reflection_id: &str, _verbose: bool) -> Result<types::ReflectionResponse> {
+        self.runtime.block_on(async {
+            let response = self.client.refresh_reflection(bank_id, reflection_id, None).await?;
+            Ok(response.into_inner())
+        })
+    }
+
+    // --- Directive Methods ---
+
+    pub fn list_directives(&self, bank_id: &str, _verbose: bool) -> Result<types::DirectiveListResponse> {
+        self.runtime.block_on(async {
+            let response = self.client.list_directives(bank_id, None, None, None, None, None, None).await?;
+            Ok(response.into_inner())
+        })
+    }
+
+    pub fn get_directive(&self, bank_id: &str, directive_id: &str, _verbose: bool) -> Result<types::DirectiveResponse> {
+        self.runtime.block_on(async {
+            let response = self.client.get_directive(bank_id, directive_id, None).await?;
+            Ok(response.into_inner())
+        })
+    }
+
+    pub fn create_directive(
+        &self,
+        bank_id: &str,
+        request: &types::CreateDirectiveRequest,
+        _verbose: bool,
+    ) -> Result<types::DirectiveResponse> {
+        self.runtime.block_on(async {
+            let response = self.client.create_directive(bank_id, None, request).await?;
+            Ok(response.into_inner())
+        })
+    }
+
+    pub fn update_directive(
+        &self,
+        bank_id: &str,
+        directive_id: &str,
+        request: &types::UpdateDirectiveRequest,
+        _verbose: bool,
+    ) -> Result<types::DirectiveResponse> {
+        self.runtime.block_on(async {
+            let response = self.client.update_directive(bank_id, directive_id, None, request).await?;
+            Ok(response.into_inner())
+        })
+    }
+
+    pub fn delete_directive(&self, bank_id: &str, directive_id: &str, _verbose: bool) -> Result<serde_json::Value> {
+        self.runtime.block_on(async {
+            let response = self.client.delete_directive(bank_id, directive_id, None).await?;
+            Ok(response.into_inner())
+        })
+    }
+}
+
 // Re-export types from the generated client for use in commands
 pub use types::{
     BankProfileResponse,

hindsight-cli/src/commands/bank.rs

@@ -222,6 +222,226 @@ pub fn update_background(
     }
 }
 
+/// Set bank mission
+pub fn mission(
+    client: &ApiClient,
+    bank_id: &str,
+    mission_text: &str,
+    verbose: bool,
+    output_format: OutputFormat,
+) -> Result<()> {
+    let spinner = if output_format == OutputFormat::Pretty {
+        Some(ui::create_spinner("Setting mission..."))
+    } else {
+        None
+    };
+
+    let response = client.set_mission(bank_id, mission_text, verbose);
+
+    if let Some(mut sp) = spinner {
+        sp.finish();
+    }
+
+    match response {
+        Ok(profile) => {
+            if output_format == OutputFormat::Pretty {
+                ui::print_success("Mission updated successfully");
+                println!();
+                println!("{}", profile.mission);
+            } else {
+                output::print_output(&profile, output_format)?;
+            }
+            Ok(())
+        }
+        Err(e) => Err(e),
+    }
+}
+
+/// Create a new bank
+pub fn create(
+    client: &ApiClient,
+    bank_id: &str,
+    name: Option<String>,
+    mission_text: Option<String>,
+    skepticism: Option<i64>,
+    literalism: Option<i64>,
+    empathy: Option<i64>,
+    verbose: bool,
+    output_format: OutputFormat,
+) -> Result<()> {
+    let spinner = if output_format == OutputFormat::Pretty {
+        Some(ui::create_spinner("Creating bank..."))
+    } else {
+        None
+    };
+
+    use hindsight_client::types;
+    use std::num::NonZeroU64;
+
+    let disposition = if skepticism.is_some() || literalism.is_some() || empathy.is_some() {
+        Some(types::DispositionTraits {
+            skepticism: NonZeroU64::new(skepticism.unwrap_or(3) as u64).unwrap(),
+            literalism: NonZeroU64::new(literalism.unwrap_or(3) as u64).unwrap(),
+            empathy: NonZeroU64::new(empathy.unwrap_or(3) as u64).unwrap(),
+        })
+    } else {
+        None
+    };
+
+    let request = types::CreateBankRequest {
+        name,
+        mission: mission_text,
+        background: None,
+        disposition,
+    };
+
+    let response = client.create_bank(bank_id, &request, verbose);
+
+    if let Some(mut sp) = spinner {
+        sp.finish();
+    }
+
+    match response {
+        Ok(profile) => {
+            if output_format == OutputFormat::Pretty {
+                ui::print_success(&format!("Bank '{}' created successfully", bank_id));
+                println!();
+                ui::print_disposition(&profile);
+            } else {
+                output::print_output(&profile, output_format)?;
+            }
+            Ok(())
+        }
+        Err(e) => Err(e),
+    }
+}
+
+/// Update bank properties (partial update)
+pub fn update(
+    client: &ApiClient,
+    bank_id: &str,
+    name: Option<String>,
+    mission_text: Option<String>,
+    skepticism: Option<i64>,
+    literalism: Option<i64>,
+    empathy: Option<i64>,
+    verbose: bool,
+    output_format: OutputFormat,
+) -> Result<()> {
+    if name.is_none() && mission_text.is_none() && skepticism.is_none() && literalism.is_none() && empathy.is_none() {
+        anyhow::bail!("At least one field must be provided (--name, --mission, --skepticism, --literalism, --empathy)");
+    }
+
+    let spinner = if output_format == OutputFormat::Pretty {
+        Some(ui::create_spinner("Updating bank..."))
+    } else {
+        None
+    };
+
+    use hindsight_client::types;
+    use std::num::NonZeroU64;
+
+    let disposition = if skepticism.is_some() || literalism.is_some() || empathy.is_some() {
+        Some(types::DispositionTraits {
+            skepticism: NonZeroU64::new(skepticism.unwrap_or(3) as u64).unwrap(),
+            literalism: NonZeroU64::new(literalism.unwrap_or(3) as u64).unwrap(),
+            empathy: NonZeroU64::new(empathy.unwrap_or(3) as u64).unwrap(),
+        })
+    } else {
+        None
+    };
+
+    let request = types::CreateBankRequest {
+        name,
+        mission: mission_text,
+        background: None,
+        disposition,
+    };
+
+    let response = client.update_bank(bank_id, &request, verbose);
+
+    if let Some(mut sp) = spinner {
+        sp.finish();
+    }
+
+    match response {
+        Ok(profile) => {
+            if output_format == OutputFormat::Pretty {
+                ui::print_success(&format!("Bank '{}' updated successfully", bank_id));
+                println!();
+                ui::print_disposition(&profile);
+            } else {
+                output::print_output(&profile, output_format)?;
+            }
+            Ok(())
+        }
+        Err(e) => Err(e),
+    }
+}
+
+/// Get memory graph data
+pub fn graph(
+    client: &ApiClient,
+    bank_id: &str,
+    type_filter: Option<String>,
+    limit: i64,
+    verbose: bool,
+    output_format: OutputFormat,
+) -> Result<()> {
+    let spinner = if output_format == OutputFormat::Pretty {
+        Some(ui::create_spinner("Fetching graph data..."))
+    } else {
+        None
+    };
+
+    let response = client.get_graph(bank_id, type_filter.as_deref(), Some(limit), verbose);
+
+    if let Some(mut sp) = spinner {
+        sp.finish();
+    }
+
+    match response {
+        Ok(result) => {
+            if output_format == OutputFormat::Pretty {
+                ui::print_section_header(&format!("Memory Graph: {}", bank_id));
+
+                println!("  {} {}", ui::dim("Nodes:"), ui::gradient_start(&result.nodes.len().to_string()));
+                println!("  {} {}", ui::dim("Edges:"), ui::gradient_end(&result.edges.len().to_string()));
+                println!();
+
+                // Show sample of nodes
+                if !result.nodes.is_empty() {
+                    println!("{}", ui::gradient_text("─── Sample Nodes ───"));
+                    for node in result.nodes.iter().take(5) {
+                        let fact_type = node.get("type")
+                            .and_then(|v| v.as_str())
+                            .unwrap_or("unknown");
+                        let id = node.get("id")
+                            .and_then(|v| v.as_str())
+                            .unwrap_or("unknown");
+                        println!("  {} [{}]", ui::dim(id), fact_type);
+                        if let Some(text) = node.get("text").and_then(|v| v.as_str()) {
+                            let preview: String = text.chars().take(60).collect();
+                            let ellipsis = if text.len() > 60 { "..." } else { "" };
+                            println!("    {}{}", preview, ellipsis);
+                        }
+                    }
+                    if result.nodes.len() > 5 {
+                        println!("  {} more...", ui::dim(&format!("+ {}", result.nodes.len() - 5)));
+                    }
+                    println!();
+                }
+
+                println!("{}", ui::dim("Use JSON output for full graph data: -o json"));
+            } else {
+                output::print_output(&result, output_format)?;
+            }
+            Ok(())
+        }
+        Err(e) => Err(e),
+    }
+}
+
 pub fn delete(
     client: &ApiClient,
     bank_id: &str,

hindsight-cli/src/commands/chunk.rs

@@ -0,0 +1,96 @@
+//! Chunk commands for retrieving document chunks.
+
+use anyhow::Result;
+
+use crate::api::ApiClient;
+use crate::output::{self, OutputFormat};
+use crate::ui;
+
+/// Get a specific chunk by ID
+pub fn get(
+    client: &ApiClient,
+    chunk_id: &str,
+    verbose: bool,
+    output_format: OutputFormat,
+) -> Result<()> {
+    let spinner = if output_format == OutputFormat::Pretty {
+        Some(ui::create_spinner("Fetching chunk..."))
+    } else {
+        None
+    };
+
+    let response = client.get_chunk(chunk_id, verbose);
+
+    if let Some(mut sp) = spinner {
+        sp.finish();
+    }
+
+    match response {
+        Ok(result) => {
+            if output_format == OutputFormat::Pretty {
+                ui::print_section_header(&format!("Chunk: {}", chunk_id));
+
+                println!("  {} {}", ui::dim("ID:"), result.chunk_id);
+                println!("  {} {}", ui::dim("Index:"), result.chunk_index);
+                println!("  {} {}", ui::dim("Document:"), result.document_id);
+                println!("  {} {}", ui::dim("Bank:"), result.bank_id);
+                println!("  {} {}", ui::dim("Created:"), result.created_at);
+
+                println!();
+                println!("{}", ui::gradient_text("─── Content ───"));
+                println!();
+                println!("{}", result.chunk_text);
+
+                println!();
+            } else {
+                output::print_output(&result, output_format)?;
+            }
+            Ok(())
+        }
+        Err(e) => Err(e),
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use hindsight_client::types::ChunkResponse;
+
+    #[test]
+    fn test_chunk_response_deserialization() {
+        let json = r#"{
+            "chunk_id": "chunk-123",
+            "bank_id": "test-bank",
+            "document_id": "doc-456",
+            "chunk_index": 0,
+            "chunk_text": "This is the chunk content.",
+            "created_at": "2024-01-15T10:00:00Z"
+        }"#;
+
+        let result: ChunkResponse = serde_json::from_str(json).unwrap();
+
+        assert_eq!(result.chunk_id, "chunk-123");
+        assert_eq!(result.bank_id, "test-bank");
+        assert_eq!(result.document_id, "doc-456");
+        assert_eq!(result.chunk_index, 0);
+        assert_eq!(result.chunk_text, "This is the chunk content.");
+        assert_eq!(result.created_at, "2024-01-15T10:00:00Z");
+    }
+
+    #[test]
+    fn test_chunk_response_multiline_content() {
+        let json = r#"{
+            "chunk_id": "chunk-456",
+            "bank_id": "test-bank",
+            "document_id": "doc-789",
+            "chunk_index": 5,
+            "chunk_text": "Line 1\nLine 2\nLine 3",
+            "created_at": "2024-01-15T11:00:00Z"
+        }"#;
+
+        let result: ChunkResponse = serde_json::from_str(json).unwrap();
+
+        assert_eq!(result.chunk_index, 5);
+        assert!(result.chunk_text.contains('\n'));
+        assert_eq!(result.chunk_text.lines().count(), 3);
+    }
+}

hindsight-cli/src/commands/directive.rs

@@ -0,0 +1,266 @@
+//! Directive commands for managing behavioral rules.
+
+use anyhow::Result;
+
+use crate::api::ApiClient;
+use crate::output::{self, OutputFormat};
+use crate::ui;
+
+use hindsight_client::types;
+
+/// List directives for a bank
+pub fn list(
+    client: &ApiClient,
+    bank_id: &str,
+    verbose: bool,
+    output_format: OutputFormat,
+) -> Result<()> {
+    let spinner = if output_format == OutputFormat::Pretty {
+        Some(ui::create_spinner("Fetching directives..."))
+    } else {
+        None
+    };
+
+    let response = client.list_directives(bank_id, verbose);
+
+    if let Some(mut sp) = spinner {
+        sp.finish();
+    }
+
+    match response {
+        Ok(result) => {
+            if output_format == OutputFormat::Pretty {
+                ui::print_section_header(&format!("Directives: {}", bank_id));
+
+                if result.items.is_empty() {
+                    println!("  {}", ui::dim("No directives found."));
+                } else {
+                    for directive in &result.items {
+                        let status = if directive.is_active {
+                            ui::gradient_start("active")
+                        } else {
+                            ui::dim("inactive")
+                        };
+                        println!(
+                            "  {} {} [{}]",
+                            ui::gradient_start(&directive.id),
+                            directive.name,
+                            status
+                        );
+
+                        // Show content preview
+                        let preview: String = directive.content.chars().take(80).collect();
+                        let ellipsis = if directive.content.len() > 80 { "..." } else { "" };
+                        println!("    {}{}", ui::dim(&preview), ellipsis);
+
+                        println!();
+                    }
+                }
+            } else {
+                output::print_output(&result, output_format)?;
+            }
+            Ok(())
+        }
+        Err(e) => Err(e),
+    }
+}
+
+/// Get a specific directive
+pub fn get(
+    client: &ApiClient,
+    bank_id: &str,
+    directive_id: &str,
+    verbose: bool,
+    output_format: OutputFormat,
+) -> Result<()> {
+    let spinner = if output_format == OutputFormat::Pretty {
+        Some(ui::create_spinner("Fetching directive..."))
+    } else {
+        None
+    };
+
+    let response = client.get_directive(bank_id, directive_id, verbose);
+
+    if let Some(mut sp) = spinner {
+        sp.finish();
+    }
+
+    match response {
+        Ok(directive) => {
+            if output_format == OutputFormat::Pretty {
+                print_directive_detail(&directive);
+            } else {
+                output::print_output(&directive, output_format)?;
+            }
+            Ok(())
+        }
+        Err(e) => Err(e),
+    }
+}
+
+/// Create a new directive
+pub fn create(
+    client: &ApiClient,
+    bank_id: &str,
+    name: &str,
+    content: &str,
+    verbose: bool,
+    output_format: OutputFormat,
+) -> Result<()> {
+    let spinner = if output_format == OutputFormat::Pretty {
+        Some(ui::create_spinner("Creating directive..."))
+    } else {
+        None
+    };
+
+    let request = types::CreateDirectiveRequest {
+        name: name.to_string(),
+        content: content.to_string(),
+        is_active: true,
+        priority: 0,
+        tags: vec![],
+    };
+
+    let response = client.create_directive(bank_id, &request, verbose);
+
+    if let Some(mut sp) = spinner {
+        sp.finish();
+    }
+
+    match response {
+        Ok(directive) => {
+            if output_format == OutputFormat::Pretty {
+                ui::print_success(&format!("Directive '{}' created successfully", directive.id));
+                println!();
+                print_directive_detail(&directive);
+            } else {
+                output::print_output(&directive, output_format)?;
+            }
+            Ok(())
+        }
+        Err(e) => Err(e),
+    }
+}
+
+/// Update a directive
+pub fn update(
+    client: &ApiClient,
+    bank_id: &str,
+    directive_id: &str,
+    name: Option<String>,
+    content: Option<String>,
+    verbose: bool,
+    output_format: OutputFormat,
+) -> Result<()> {
+    if name.is_none() && content.is_none() {
+        anyhow::bail!("At least one of --name or --content must be provided");
+    }
+
+    let spinner = if output_format == OutputFormat::Pretty {
+        Some(ui::create_spinner("Updating directive..."))
+    } else {
+        None
+    };
+
+    let request = types::UpdateDirectiveRequest {
+        name,
+        content,
+        is_active: None,
+        priority: None,
+        tags: None,
+    };
+
+    let response = client.update_directive(bank_id, directive_id, &request, verbose);
+
+    if let Some(mut sp) = spinner {
+        sp.finish();
+    }
+
+    match response {
+        Ok(directive) => {
+            if output_format == OutputFormat::Pretty {
+                ui::print_success(&format!("Directive '{}' updated successfully", directive_id));
+                println!();
+                print_directive_detail(&directive);
+            } else {
+                output::print_output(&directive, output_format)?;
+            }
+            Ok(())
+        }
+        Err(e) => Err(e),
+    }
+}
+
+/// Delete a directive
+pub fn delete(
+    client: &ApiClient,
+    bank_id: &str,
+    directive_id: &str,
+    yes: bool,
+    verbose: bool,
+    output_format: OutputFormat,
+) -> Result<()> {
+    // Confirmation prompt unless -y flag is used
+    if !yes && output_format == OutputFormat::Pretty {
+        let message = format!(
+            "Are you sure you want to delete directive '{}'? This cannot be undone.",
+            directive_id
+        );
+
+        let confirmed = ui::prompt_confirmation(&message)?;
+
+        if !confirmed {
+            ui::print_info("Operation cancelled");
+            return Ok(());
+        }
+    }
+
+    let spinner = if output_format == OutputFormat::Pretty {
+        Some(ui::create_spinner("Deleting directive..."))
+    } else {
+        None
+    };
+
+    let response = client.delete_directive(bank_id, directive_id, verbose);
+
+    if let Some(mut sp) = spinner {
+        sp.finish();
+    }
+
+    match response {
+        Ok(_) => {
+            if output_format == OutputFormat::Pretty {
+                ui::print_success(&format!("Directive '{}' deleted successfully", directive_id));
+            } else {
+                println!("{{\"success\": true}}");
+            }
+            Ok(())
+        }
+        Err(e) => Err(e),
+    }
+}
+
+// Helper function to print directive details
+fn print_directive_detail(directive: &types::DirectiveResponse) {
+    ui::print_section_header(&directive.name);
+
+    println!("  {} {}", ui::dim("ID:"), ui::gradient_start(&directive.id));
+
+    let status = if directive.is_active {
+        ui::gradient_start("active")
+    } else {
+        ui::dim("inactive")
+    };
+    println!("  {} {}", ui::dim("Status:"), status);
+    println!("  {} {}", ui::dim("Priority:"), directive.priority);
+
+    if !directive.tags.is_empty() {
+        println!("  {} {}", ui::dim("Tags:"), directive.tags.join(", "));
+    }
+
+    println!();
+    println!("{}", ui::gradient_text("─── Content ───"));
+    println!();
+    println!("{}", &directive.content);
+    println!();
+}

hindsight-cli/src/commands/health.rs

@@ -0,0 +1,157 @@
+//! Health and metrics commands.
+
+use anyhow::Result;
+use serde::Deserialize;
+
+use crate::api::ApiClient;
+use crate::output::{self, OutputFormat};
+use crate::ui;
+
+// Local type for health response
+#[derive(Debug, Deserialize)]
+struct HealthResponse {
+    status: String,
+    database: Option<String>,
+    version: Option<String>,
+}
+
+/// Check API health
+pub fn health(
+    client: &ApiClient,
+    verbose: bool,
+    output_format: OutputFormat,
+) -> Result<()> {
+    let spinner = if output_format == OutputFormat::Pretty {
+        Some(ui::create_spinner("Checking health..."))
+    } else {
+        None
+    };
+
+    let response = client.health(verbose);
+
+    if let Some(mut sp) = spinner {
+        sp.finish();
+    }
+
+    match response {
+        Ok(value) => {
+            if output_format == OutputFormat::Pretty {
+                let result: HealthResponse = serde_json::from_value(value.clone())
+                    .unwrap_or(HealthResponse {
+                        status: "unknown".to_string(),
+                        database: None,
+                        version: None,
+                    });
+
+                let status_str = if result.status == "healthy" {
+                    ui::gradient_start(&result.status)
+                } else {
+                    ui::gradient_end(&result.status)
+                };
+
+                ui::print_section_header("Health Check");
+                println!("  {} {}", ui::dim("Status:"), status_str);
+
+                if let Some(db_status) = &result.database {
+                    let db_str = if db_status == "connected" {
+                        ui::gradient_start(db_status)
+                    } else {
+                        ui::gradient_end(db_status)
+                    };
+                    println!("  {} {}", ui::dim("Database:"), db_str);
+                }
+
+                if let Some(version) = &result.version {
+                    println!("  {} {}", ui::dim("Version:"), version);
+                }
+
+                println!();
+            } else {
+                output::print_output(&value, output_format)?;
+            }
+            Ok(())
+        }
+        Err(e) => Err(e),
+    }
+}
+
+/// Get Prometheus metrics
+pub fn metrics(
+    client: &ApiClient,
+    verbose: bool,
+    output_format: OutputFormat,
+) -> Result<()> {
+    let spinner = if output_format == OutputFormat::Pretty {
+        Some(ui::create_spinner("Fetching metrics..."))
+    } else {
+        None
+    };
+
+    let response = client.metrics(verbose);
+
+    if let Some(mut sp) = spinner {
+        sp.finish();
+    }
+
+    match response {
+        Ok(result) => {
+            if output_format == OutputFormat::Pretty {
+                ui::print_section_header("Prometheus Metrics");
+                println!("{}", result);
+            } else {
+                // For JSON/YAML, wrap in an object
+                let wrapped = serde_json::json!({ "metrics": result });
+                output::print_output(&wrapped, output_format)?;
+            }
+            Ok(())
+        }
+        Err(e) => Err(e),
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_health_response_deserialization() {
+        let json = r#"{
+            "status": "healthy",
+            "database": "connected",
+            "version": "0.3.0"
+        }"#;
+
+        let value: serde_json::Value = serde_json::from_str(json).unwrap();
+        let result: HealthResponse = serde_json::from_value(value).unwrap();
+
+        assert_eq!(result.status, "healthy");
+        assert_eq!(result.database, Some("connected".to_string()));
+        assert_eq!(result.version, Some("0.3.0".to_string()));
+    }
+
+    #[test]
+    fn test_health_response_minimal() {
+        let json = r#"{"status": "healthy"}"#;
+
+        let value: serde_json::Value = serde_json::from_str(json).unwrap();
+        let result: HealthResponse = serde_json::from_value(value).unwrap();
+
+        assert_eq!(result.status, "healthy");
+        assert_eq!(result.database, None);
+        assert_eq!(result.version, None);
+    }
+
+    #[test]
+    fn test_health_response_unhealthy() {
+        let json = r#"{
+            "status": "unhealthy",
+            "database": "disconnected"
+        }"#;
+
+        let value: serde_json::Value = serde_json::from_str(json).unwrap();
+        let result: HealthResponse = serde_json::from_value(value).unwrap();
+
+        assert_eq!(result.status, "unhealthy");
+        assert_eq!(result.database, Some("disconnected".to_string()));
+    }
+}

hindsight-cli/src/commands/memory.rs

@@ -10,8 +10,30 @@ use crate::ui;
 
 // Import types from generated client
 use hindsight_client::types::{Budget, ChunkIncludeOptions, IncludeOptions, TagsMatch};
+use serde::Deserialize;
 use serde_json;
 
+// Local types for serde_json::Value deserialization
+#[derive(Debug, Deserialize)]
+struct MemoryUnitDetail {
+    id: String,
+    text: String,
+    #[serde(rename = "type")]
+    type_: Option<String>,
+    document_id: Option<String>,
+    context: Option<String>,
+    occurred_start: Option<String>,
+    occurred_end: Option<String>,
+    entities: Option<Vec<EntityRef>>,
+    tags: Option<Vec<String>>,
+}
+
+#[derive(Debug, Deserialize)]
+struct EntityRef {
+    id: String,
+    name: String,
+}
+
 // Helper function to parse budget string to Budget enum
 fn parse_budget(budget: &str) -> Budget {
     match budget.to_lowercase().as_str() {
@@ -21,6 +43,183 @@ fn parse_budget(budget: &str) -> Budget {
     }
 }
 
+/// List memory units with pagination and optional filters
+pub fn list(
+    client: &ApiClient,
+    bank_id: &str,
+    type_filter: Option<String>,
+    query: Option<String>,
+    limit: i64,
+    offset: i64,
+    verbose: bool,
+    output_format: OutputFormat,
+) -> Result<()> {
+    let spinner = if output_format == OutputFormat::Pretty {
+        Some(ui::create_spinner("Fetching memories..."))
+    } else {
+        None
+    };
+
+    let response = client.list_memories(
+        bank_id,
+        type_filter.as_deref(),
+        query.as_deref(),
+        Some(limit),
+        Some(offset),
+        verbose,
+    );
+
+    if let Some(mut sp) = spinner {
+        sp.finish();
+    }
+
+    match response {
+        Ok(result) => {
+            if output_format == OutputFormat::Pretty {
+                ui::print_section_header(&format!("Memories: {} (showing {}-{})", bank_id, offset + 1, offset + result.items.len() as i64));
+
+                if result.items.is_empty() {
+                    println!("  {}", ui::dim("No memories found."));
+                } else {
+                    for item in &result.items {
+                        let fact_type = item.get("type")
+                            .and_then(|v| v.as_str())
+                            .unwrap_or("unknown");
+                        let type_t = match fact_type {
+                            "world" => 0.0,
+                            "experience" => 0.5,
+                            "opinion" => 1.0,
+                            _ => 0.5,
+                        };
+
+                        let id = item.get("id")
+                            .and_then(|v| v.as_str())
+                            .unwrap_or("unknown");
+
+                        println!(
+                            "  {} {}",
+                            ui::gradient(&format!("[{}]", fact_type.to_uppercase()), type_t),
+                            ui::dim(id)
+                        );
+
+                        // Truncate text if too long
+                        if let Some(text) = item.get("text").and_then(|v| v.as_str()) {
+                            let text_preview: String = text.chars().take(100).collect();
+                            let ellipsis = if text.len() > 100 { "..." } else { "" };
+                            println!("    {}{}", text_preview, ellipsis);
+                        }
+
+                        if let Some(doc_id) = item.get("document_id").and_then(|v| v.as_str()) {
+                            println!("    {} {}", ui::dim("doc:"), ui::dim(doc_id));
+                        }
+                        println!();
+                    }
+
+                    println!("  {} {} total", ui::dim("Total:"), result.total);
+                }
+            } else {
+                output::print_output(&result, output_format)?;
+            }
+            Ok(())
+        }
+        Err(e) => Err(e),
+    }
+}
+
+/// Get a specific memory unit by ID
+pub fn get(
+    client: &ApiClient,
+    bank_id: &str,
+    memory_id: &str,
+    verbose: bool,
+    output_format: OutputFormat,
+) -> Result<()> {
+    let spinner = if output_format == OutputFormat::Pretty {
+        Some(ui::create_spinner("Fetching memory..."))
+    } else {
+        None
+    };
+
+    let response = client.get_memory(bank_id, memory_id, verbose);
+
+    if let Some(mut sp) = spinner {
+        sp.finish();
+    }
+
+    match response {
+        Ok(value) => {
+            if output_format == OutputFormat::Pretty {
+                let result: MemoryUnitDetail = serde_json::from_value(value)
+                    .with_context(|| "Failed to parse memory response")?;
+
+                let fact_type = result.type_.as_deref().unwrap_or("unknown");
+                let type_t = match fact_type {
+                    "world" => 0.0,
+                    "experience" => 0.5,
+                    "opinion" => 1.0,
+                    _ => 0.5,
+                };
+
+                ui::print_section_header(&format!("Memory: {}", memory_id));
+
+                println!("  {} {}", ui::dim("Type:"), ui::gradient(&fact_type.to_uppercase(), type_t));
+                println!("  {} {}", ui::dim("ID:"), result.id);
+
+                if let Some(doc_id) = &result.document_id {
+                    println!("  {} {}", ui::dim("Document:"), doc_id);
+                }
+
+                if let Some(context) = &result.context {
+                    println!("  {} {}", ui::dim("Context:"), context);
+                }
+
+                println!();
+                println!("{}", ui::gradient_text("─── Content ───"));
+                println!();
+                println!("{}", result.text);
+
+                // Show temporal info if available
+                if result.occurred_start.is_some() || result.occurred_end.is_some() {
+                    println!();
+                    println!("{}", ui::gradient_text("─── Temporal ───"));
+                    if let Some(start) = &result.occurred_start {
+                        println!("  {} {}", ui::dim("Start:"), start);
+                    }
+                    if let Some(end) = &result.occurred_end {
+                        println!("  {} {}", ui::dim("End:"), end);
+                    }
+                }
+
+                // Show entities if available
+                if let Some(entities) = &result.entities {
+                    if !entities.is_empty() {
+                        println!();
+                        println!("{}", ui::gradient_text("─── Entities ───"));
+                        for entity in entities {
+                            println!("  • {} ({})", entity.name, entity.id);
+                        }
+                    }
+                }
+
+                // Show tags if available
+                if let Some(tags) = &result.tags {
+                    if !tags.is_empty() {
+                        println!();
+                        println!("{}", ui::gradient_text("─── Tags ───"));
+                        println!("  {}", tags.join(", "));
+                    }
+                }
+
+                println!();
+            } else {
+                output::print_output(&value, output_format)?;
+            }
+            Ok(())
+        }
+        Err(e) => Err(e),
+    }
+}
+
 // Helper function to check if a file has a text-based extension
 fn is_text_file(path: &std::path::Path) -> bool {
     const TEXT_EXTENSIONS: &[&str] = &[

hindsight-cli/src/commands/mod.rs

@@ -1,6 +1,11 @@
 pub mod bank;
-pub mod memory;
+pub mod chunk;
+pub mod directive;
 pub mod document;
 pub mod entity;
-pub mod operation;
 pub mod explore;
+pub mod health;
+pub mod memory;
+pub mod operation;
+pub mod reflection;
+pub mod tag;

hindsight-cli/src/commands/operation.rs

@@ -47,6 +47,55 @@ pub fn list(
     }
 }
 
+/// Get the status of a specific operation
+pub fn get(
+    client: &ApiClient,
+    agent_id: &str,
+    operation_id: &str,
+    verbose: bool,
+    output_format: OutputFormat,
+) -> Result<()> {
+    let spinner = if output_format == OutputFormat::Pretty {
+        Some(ui::create_spinner("Fetching operation status..."))
+    } else {
+        None
+    };
+
+    let response = client.get_operation(agent_id, operation_id, verbose);
+
+    if let Some(mut sp) = spinner {
+        sp.finish();
+    }
+
+    match response {
+        Ok(result) => {
+            if output_format == OutputFormat::Pretty {
+                ui::print_section_header(&format!("Operation: {}", operation_id));
+
+                use hindsight_client::types::Status;
+                let status_str = match &result.status {
+                    Status::Completed => ui::gradient_start("completed"),
+                    Status::Pending => ui::gradient_mid("pending"),
+                    Status::Failed => ui::gradient_end("failed"),
+                    Status::NotFound => ui::gradient_end("not_found"),
+                };
+
+                println!("  {} {}", ui::dim("Status:"), status_str);
+
+                if let Some(error) = &result.error_message {
+                    println!("  {} {}", ui::dim("Error:"), ui::gradient_end(error));
+                }
+
+                println!();
+            } else {
+                output::print_output(&result, output_format)?;
+            }
+            Ok(())
+        }
+        Err(e) => Err(e),
+    }
+}
+
 pub fn cancel(
     client: &ApiClient,
     agent_id: &str,

hindsight-cli/src/commands/reflection.rs

@@ -0,0 +1,274 @@
+//! Reflection commands for managing user-curated summaries.
+
+use anyhow::Result;
+
+use crate::api::ApiClient;
+use crate::output::{self, OutputFormat};
+use crate::ui;
+
+use hindsight_client::types;
+
+/// List reflections for a bank
+pub fn list(
+    client: &ApiClient,
+    bank_id: &str,
+    verbose: bool,
+    output_format: OutputFormat,
+) -> Result<()> {
+    let spinner = if output_format == OutputFormat::Pretty {
+        Some(ui::create_spinner("Fetching reflections..."))
+    } else {
+        None
+    };
+
+    let response = client.list_reflections(bank_id, verbose);
+
+    if let Some(mut sp) = spinner {
+        sp.finish();
+    }
+
+    match response {
+        Ok(result) => {
+            if output_format == OutputFormat::Pretty {
+                ui::print_section_header(&format!("Reflections: {}", bank_id));
+
+                if result.items.is_empty() {
+                    println!("  {}", ui::dim("No reflections found."));
+                } else {
+                    for reflection in &result.items {
+                        println!(
+                            "  {} {}",
+                            ui::gradient_start(&reflection.id),
+                            reflection.name
+                        );
+
+                        // Show content preview
+                        let preview: String = reflection.content.chars().take(80).collect();
+                        let ellipsis = if reflection.content.len() > 80 { "..." } else { "" };
+                        println!("    {}{}", ui::dim(&preview), ellipsis);
+
+                        println!();
+                    }
+                }
+            } else {
+                output::print_output(&result, output_format)?;
+            }
+            Ok(())
+        }
+        Err(e) => Err(e),
+    }
+}
+
+/// Get a specific reflection
+pub fn get(
+    client: &ApiClient,
+    bank_id: &str,
+    reflection_id: &str,
+    verbose: bool,
+    output_format: OutputFormat,
+) -> Result<()> {
+    let spinner = if output_format == OutputFormat::Pretty {
+        Some(ui::create_spinner("Fetching reflection..."))
+    } else {
+        None
+    };
+
+    let response = client.get_reflection(bank_id, reflection_id, verbose);
+
+    if let Some(mut sp) = spinner {
+        sp.finish();
+    }
+
+    match response {
+        Ok(reflection) => {
+            if output_format == OutputFormat::Pretty {
+                print_reflection_detail(&reflection);
+            } else {
+                output::print_output(&reflection, output_format)?;
+            }
+            Ok(())
+        }
+        Err(e) => Err(e),
+    }
+}
+
+/// Create a new reflection
+pub fn create(
+    client: &ApiClient,
+    bank_id: &str,
+    name: &str,
+    source_query: &str,
+    verbose: bool,
+    output_format: OutputFormat,
+) -> Result<()> {
+    let spinner = if output_format == OutputFormat::Pretty {
+        Some(ui::create_spinner("Creating reflection..."))
+    } else {
+        None
+    };
+
+    let request = types::CreateReflectionRequest {
+        name: name.to_string(),
+        source_query: source_query.to_string(),
+        max_tokens: 2048,
+        tags: vec![],
+    };
+
+    let response = client.create_reflection(bank_id, &request, verbose);
+
+    if let Some(mut sp) = spinner {
+        sp.finish();
+    }
+
+    match response {
+        Ok(result) => {
+            if output_format == OutputFormat::Pretty {
+                ui::print_success(&format!("Reflection created, operation_id: {}", result.operation_id));
+            } else {
+                output::print_output(&result, output_format)?;
+            }
+            Ok(())
+        }
+        Err(e) => Err(e),
+    }
+}
+
+/// Update a reflection
+pub fn update(
+    client: &ApiClient,
+    bank_id: &str,
+    reflection_id: &str,
+    name: Option<String>,
+    verbose: bool,
+    output_format: OutputFormat,
+) -> Result<()> {
+    if name.is_none() {
+        anyhow::bail!("--name must be provided");
+    }
+
+    let spinner = if output_format == OutputFormat::Pretty {
+        Some(ui::create_spinner("Updating reflection..."))
+    } else {
+        None
+    };
+
+    let request = types::UpdateReflectionRequest { name };
+
+    let response = client.update_reflection(bank_id, reflection_id, &request, verbose);
+
+    if let Some(mut sp) = spinner {
+        sp.finish();
+    }
+
+    match response {
+        Ok(reflection) => {
+            if output_format == OutputFormat::Pretty {
+                ui::print_success(&format!("Reflection '{}' updated successfully", reflection_id));
+                println!();
+                print_reflection_detail(&reflection);
+            } else {
+                output::print_output(&reflection, output_format)?;
+            }
+            Ok(())
+        }
+        Err(e) => Err(e),
+    }
+}
+
+/// Delete a reflection
+pub fn delete(
+    client: &ApiClient,
+    bank_id: &str,
+    reflection_id: &str,
+    yes: bool,
+    verbose: bool,
+    output_format: OutputFormat,
+) -> Result<()> {
+    // Confirmation prompt unless -y flag is used
+    if !yes && output_format == OutputFormat::Pretty {
+        let message = format!(
+            "Are you sure you want to delete reflection '{}'? This cannot be undone.",
+            reflection_id
+        );
+
+        let confirmed = ui::prompt_confirmation(&message)?;
+
+        if !confirmed {
+            ui::print_info("Operation cancelled");
+            return Ok(());
+        }
+    }
+
+    let spinner = if output_format == OutputFormat::Pretty {
+        Some(ui::create_spinner("Deleting reflection..."))
+    } else {
+        None
+    };
+
+    let response = client.delete_reflection(bank_id, reflection_id, verbose);
+
+    if let Some(mut sp) = spinner {
+        sp.finish();
+    }
+
+    match response {
+        Ok(_) => {
+            if output_format == OutputFormat::Pretty {
+                ui::print_success(&format!("Reflection '{}' deleted successfully", reflection_id));
+            } else {
+                println!("{{\"success\": true}}");
+            }
+            Ok(())
+        }
+        Err(e) => Err(e),
+    }
+}
+
+/// Refresh a reflection
+pub fn refresh(
+    client: &ApiClient,
+    bank_id: &str,
+    reflection_id: &str,
+    verbose: bool,
+    output_format: OutputFormat,
+) -> Result<()> {
+    let spinner = if output_format == OutputFormat::Pretty {
+        Some(ui::create_spinner("Refreshing reflection..."))
+    } else {
+        None
+    };
+
+    let response = client.refresh_reflection(bank_id, reflection_id, verbose);
+
+    if let Some(mut sp) = spinner {
+        sp.finish();
+    }
+
+    match response {
+        Ok(reflection) => {
+            if output_format == OutputFormat::Pretty {
+                ui::print_success(&format!("Reflection '{}' refreshed successfully", reflection_id));
+                println!();
+                print_reflection_detail(&reflection);
+            } else {
+                output::print_output(&reflection, output_format)?;
+            }
+            Ok(())
+        }
+        Err(e) => Err(e),
+    }
+}
+
+// Helper function to print reflection details
+fn print_reflection_detail(reflection: &types::ReflectionResponse) {
+    ui::print_section_header(&reflection.name);
+
+    println!("  {} {}", ui::dim("ID:"), ui::gradient_start(&reflection.id));
+    println!("  {} {}", ui::dim("Source Query:"), &reflection.source_query);
+
+    println!();
+    println!("{}", ui::gradient_text("─── Content ───"));
+    println!();
+    println!("{}", &reflection.content);
+    println!();
+}

hindsight-cli/src/commands/tag.rs

@@ -0,0 +1,119 @@
+//! Tag commands for listing tags in a memory bank.
+
+use anyhow::Result;
+
+use crate::api::ApiClient;
+use crate::output::{self, OutputFormat};
+use crate::ui;
+
+/// List tags in a bank
+pub fn list(
+    client: &ApiClient,
+    bank_id: &str,
+    query: Option<String>,
+    limit: i64,
+    offset: i64,
+    verbose: bool,
+    output_format: OutputFormat,
+) -> Result<()> {
+    let spinner = if output_format == OutputFormat::Pretty {
+        Some(ui::create_spinner("Fetching tags..."))
+    } else {
+        None
+    };
+
+    let response = client.list_tags(
+        bank_id,
+        query.as_deref(),
+        Some(limit),
+        Some(offset),
+        verbose,
+    );
+
+    if let Some(mut sp) = spinner {
+        sp.finish();
+    }
+
+    match response {
+        Ok(result) => {
+            if output_format == OutputFormat::Pretty {
+                ui::print_section_header(&format!("Tags: {}", bank_id));
+
+                if result.items.is_empty() {
+                    println!("  {}", ui::dim("No tags found."));
+                } else {
+                    for (i, tag) in result.items.iter().enumerate() {
+                        let t = i as f32 / result.items.len().max(1) as f32;
+                        println!(
+                            "  {} {}",
+                            ui::gradient(&tag.tag, t),
+                            ui::dim(&format!("({})", tag.count))
+                        );
+                    }
+                    println!();
+                    println!("  {} {} total", ui::dim("Total:"), result.total);
+                }
+            } else {
+                output::print_output(&result, output_format)?;
+            }
+            Ok(())
+        }
+        Err(e) => Err(e),
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use hindsight_client::types::{ListTagsResponse, TagItem};
+
+    #[test]
+    fn test_tag_item_fields() {
+        // Verify TagItem has the expected fields
+        let tag = TagItem {
+            tag: "test-tag".to_string(),
+            count: 5,
+        };
+
+        assert_eq!(tag.tag, "test-tag");
+        assert_eq!(tag.count, 5);
+    }
+
+    #[test]
+    fn test_list_tags_response_deserialization() {
+        let json = r#"{
+            "items": [
+                {"tag": "user", "count": 10},
+                {"tag": "system", "count": 5}
+            ],
+            "limit": 100,
+            "offset": 0,
+            "total": 2
+        }"#;
+
+        let result: ListTagsResponse = serde_json::from_str(json).unwrap();
+
+        assert_eq!(result.items.len(), 2);
+        assert_eq!(result.items[0].tag, "user");
+        assert_eq!(result.items[0].count, 10);
+        assert_eq!(result.items[1].tag, "system");
+        assert_eq!(result.items[1].count, 5);
+        assert_eq!(result.total, 2);
+        assert_eq!(result.limit, 100);
+        assert_eq!(result.offset, 0);
+    }
+
+    #[test]
+    fn test_empty_tags_response() {
+        let json = r#"{
+            "items": [],
+            "limit": 100,
+            "offset": 0,
+            "total": 0
+        }"#;
+
+        let result: ListTagsResponse = serde_json::from_str(json).unwrap();
+
+        assert!(result.items.is_empty());
+        assert_eq!(result.total, 0);
+    }
+}

hindsight-cli/src/main.rs

@@ -67,11 +67,11 @@ fn get_before_help() -> &'static str {
 
 #[derive(Subcommand)]
 enum Commands {
-    /// Manage banks (list, profile, stats)
+    /// Manage banks (list, create, update, profile, stats, mission, graph, delete)
     #[command(subcommand)]
     Bank(BankCommands),
 
-    /// Manage memories (recall, reflect, retain, delete)
+    /// Manage memories (list, get, recall, reflect, retain, clear)
     #[command(subcommand)]
     Memory(MemoryCommands),
 
@@ -83,10 +83,32 @@ enum Commands {
     #[command(subcommand)]
     Entity(EntityCommands),
 
-    /// Manage async operations (list, cancel)
+    /// Manage tags (list)
+    #[command(subcommand)]
+    Tag(TagCommands),
+
+    /// Manage chunks (get)
+    #[command(subcommand)]
+    Chunk(ChunkCommands),
+
+    /// Manage async operations (list, get, cancel)
     #[command(subcommand)]
     Operation(OperationCommands),
 
+    /// Manage reflections (user-curated summaries)
+    #[command(subcommand)]
+    Reflection(ReflectionCommands),
+
+    /// Manage directives (behavioral rules)
+    #[command(subcommand)]
+    Directive(DirectiveCommands),
+
+    /// Check API health status
+    Health,
+
+    /// Get Prometheus metrics
+    Metrics,
+
     /// Interactive TUI explorer (k9s-style) for navigating banks, memories, entities, and performing recall/reflect
     #[command(alias = "tui")]
     Explore,
@@ -111,7 +133,59 @@ enum BankCommands {
     /// List all banks
     List,
 
-    /// Get bank disposition and background
+    /// Create a new bank
+    Create {
+        /// Bank ID
+        bank_id: String,
+
+        /// Bank name
+        #[arg(short = 'n', long)]
+        name: Option<String>,
+
+        /// Mission statement
+        #[arg(short = 'm', long)]
+        mission: Option<String>,
+
+        /// Skepticism trait (1-5)
+        #[arg(long, value_parser = clap::value_parser!(i64).range(1..=5))]
+        skepticism: Option<i64>,
+
+        /// Literalism trait (1-5)
+        #[arg(long, value_parser = clap::value_parser!(i64).range(1..=5))]
+        literalism: Option<i64>,
+
+        /// Empathy trait (1-5)
+        #[arg(long, value_parser = clap::value_parser!(i64).range(1..=5))]
+        empathy: Option<i64>,
+    },
+
+    /// Update bank properties (partial update)
+    Update {
+        /// Bank ID
+        bank_id: String,
+
+        /// Bank name
+        #[arg(short = 'n', long)]
+        name: Option<String>,
+
+        /// Mission statement
+        #[arg(short = 'm', long)]
+        mission: Option<String>,
+
+        /// Skepticism trait (1-5)
+        #[arg(long, value_parser = clap::value_parser!(i64).range(1..=5))]
+        skepticism: Option<i64>,
+
+        /// Literalism trait (1-5)
+        #[arg(long, value_parser = clap::value_parser!(i64).range(1..=5))]
+        literalism: Option<i64>,
+
+        /// Empathy trait (1-5)
+        #[arg(long, value_parser = clap::value_parser!(i64).range(1..=5))]
+        empathy: Option<i64>,
+    },
+
+    /// Get bank disposition and profile
     Disposition {
         /// Bank ID
         bank_id: String,
@@ -132,7 +206,17 @@ enum BankCommands {
         name: String,
     },
 
-    /// Set or merge bank background
+    /// Set bank mission
+    Mission {
+        /// Bank ID
+        bank_id: String,
+
+        /// Mission statement
+        mission: String,
+    },
+
+    /// Set or merge bank background (deprecated: use mission instead)
+    #[command(hide = true)]
     Background {
         /// Bank ID
         bank_id: String,
@@ -145,6 +229,20 @@ enum BankCommands {
         no_update_disposition: bool,
     },
 
+    /// Get memory graph data
+    Graph {
+        /// Bank ID
+        bank_id: String,
+
+        /// Filter by fact type (world, experience, opinion)
+        #[arg(short = 't', long)]
+        fact_type: Option<String>,
+
+        /// Maximum nodes to return
+        #[arg(short = 'l', long, default_value = "1000")]
+        limit: i64,
+    },
+
     /// Delete a bank and all its data
     Delete {
         /// Bank ID
@@ -158,6 +256,37 @@ enum BankCommands {
 
 #[derive(Subcommand)]
 enum MemoryCommands {
+    /// List memory units with pagination
+    List {
+        /// Bank ID
+        bank_id: String,
+
+        /// Filter by fact type (world, experience, opinion)
+        #[arg(short = 't', long)]
+        fact_type: Option<String>,
+
+        /// Full-text search query
+        #[arg(short = 'q', long)]
+        query: Option<String>,
+
+        /// Maximum number of results
+        #[arg(short = 'l', long, default_value = "100")]
+        limit: i64,
+
+        /// Offset for pagination
+        #[arg(short = 's', long, default_value = "0")]
+        offset: i64,
+    },
+
+    /// Get a specific memory unit by ID
+    Get {
+        /// Bank ID
+        bank_id: String,
+
+        /// Memory unit ID
+        memory_id: String,
+    },
+
     /// Recall memories using semantic search
     Recall {
         /// Bank ID
@@ -360,6 +489,15 @@ enum OperationCommands {
         bank_id: String,
     },
 
+    /// Get the status of a specific operation
+    Get {
+        /// Bank ID
+        bank_id: String,
+
+        /// Operation ID
+        operation_id: String,
+    },
+
     /// Cancel a pending async operation
     Cancel {
         /// Bank ID
@@ -370,6 +508,161 @@ enum OperationCommands {
     },
 }
 
+#[derive(Subcommand)]
+enum TagCommands {
+    /// List tags in a bank
+    List {
+        /// Bank ID
+        bank_id: String,
+
+        /// Wildcard search query (e.g., 'user:*')
+        #[arg(short = 'q', long)]
+        query: Option<String>,
+
+        /// Maximum number of results
+        #[arg(short = 'l', long, default_value = "100")]
+        limit: i64,
+
+        /// Offset for pagination
+        #[arg(short = 's', long, default_value = "0")]
+        offset: i64,
+    },
+}
+
+#[derive(Subcommand)]
+enum ChunkCommands {
+    /// Get a specific chunk by ID
+    Get {
+        /// Chunk ID
+        chunk_id: String,
+    },
+}
+
+#[derive(Subcommand)]
+enum ReflectionCommands {
+    /// List reflections for a bank
+    List {
+        /// Bank ID
+        bank_id: String,
+    },
+
+    /// Get a specific reflection
+    Get {
+        /// Bank ID
+        bank_id: String,
+
+        /// Reflection ID
+        reflection_id: String,
+    },
+
+    /// Create a new reflection
+    Create {
+        /// Bank ID
+        bank_id: String,
+
+        /// Reflection name
+        name: String,
+
+        /// Source query to generate the reflection from
+        source_query: String,
+    },
+
+    /// Update a reflection
+    Update {
+        /// Bank ID
+        bank_id: String,
+
+        /// Reflection ID
+        reflection_id: String,
+
+        /// New name
+        #[arg(long)]
+        name: Option<String>,
+    },
+
+    /// Delete a reflection
+    Delete {
+        /// Bank ID
+        bank_id: String,
+
+        /// Reflection ID
+        reflection_id: String,
+
+        /// Skip confirmation prompt
+        #[arg(short = 'y', long)]
+        yes: bool,
+    },
+
+    /// Refresh a reflection (re-run the source query)
+    Refresh {
+        /// Bank ID
+        bank_id: String,
+
+        /// Reflection ID
+        reflection_id: String,
+    },
+}
+
+#[derive(Subcommand)]
+enum DirectiveCommands {
+    /// List directives for a bank
+    List {
+        /// Bank ID
+        bank_id: String,
+    },
+
+    /// Get a specific directive
+    Get {
+        /// Bank ID
+        bank_id: String,
+
+        /// Directive ID
+        directive_id: String,
+    },
+
+    /// Create a new directive
+    Create {
+        /// Bank ID
+        bank_id: String,
+
+        /// Directive name
+        name: String,
+
+        /// Directive content (the text to inject into prompts)
+        content: String,
+    },
+
+    /// Update a directive
+    Update {
+        /// Bank ID
+        bank_id: String,
+
+        /// Directive ID
+        directive_id: String,
+
+        /// New name
+        #[arg(long)]
+        name: Option<String>,
+
+        /// New content
+        #[arg(long)]
+        content: Option<String>,
+    },
+
+    /// Delete a directive
+    Delete {
+        /// Bank ID
+        bank_id: String,
+
+        /// Directive ID
+        directive_id: String,
+
+        /// Skip confirmation prompt
+        #[arg(short = 'y', long)]
+        yes: bool,
+    },
+}
+
 fn main() {
     if let Err(_) = run() {
         std::process::exit(1);
@@ -412,20 +705,45 @@ fn run() -> Result<()> {
         Commands::Configure { .. } => unreachable!(), // Handled above
         Commands::Ui => unreachable!(), // Handled above
         Commands::Explore => commands::explore::run(&client),
+
+        // Health and Metrics
+        Commands::Health => commands::health::health(&client, verbose, output_format),
+        Commands::Metrics => commands::health::metrics(&client, verbose, output_format),
+
+        // Bank commands
         Commands::Bank(bank_cmd) => match bank_cmd {
             BankCommands::List => commands::bank::list(&client, verbose, output_format),
+            BankCommands::Create { bank_id, name, mission, skepticism, literalism, empathy } => {
+                commands::bank::create(&client, &bank_id, name, mission, skepticism, literalism, empathy, verbose, output_format)
+            }
+            BankCommands::Update { bank_id, name, mission, skepticism, literalism, empathy } => {
+                commands::bank::update(&client, &bank_id, name, mission, skepticism, literalism, empathy, verbose, output_format)
+            }
             BankCommands::Disposition { bank_id } => commands::bank::disposition(&client, &bank_id, verbose, output_format),
             BankCommands::Stats { bank_id } => commands::bank::stats(&client, &bank_id, verbose, output_format),
             BankCommands::Name { bank_id, name } => commands::bank::update_name(&client, &bank_id, &name, verbose, output_format),
+            BankCommands::Mission { bank_id, mission } => {
+                commands::bank::mission(&client, &bank_id, &mission, verbose, output_format)
+            }
             BankCommands::Background { bank_id, content, no_update_disposition } => {
                 commands::bank::update_background(&client, &bank_id, &content, no_update_disposition, verbose, output_format)
             }
+            BankCommands::Graph { bank_id, fact_type, limit } => {
+                commands::bank::graph(&client, &bank_id, fact_type, limit, verbose, output_format)
+            }
             BankCommands::Delete { bank_id, yes } => {
                 commands::bank::delete(&client, &bank_id, yes, verbose, output_format)
             }
         },
 
+        // Memory commands
         Commands::Memory(memory_cmd) => match memory_cmd {
+            MemoryCommands::List { bank_id, fact_type, query, limit, offset } => {
+                commands::memory::list(&client, &bank_id, fact_type, query, limit, offset, verbose, output_format)
+            }
+            MemoryCommands::Get { bank_id, memory_id } => {
+                commands::memory::get(&client, &bank_id, &memory_id, verbose, output_format)
+            }
             MemoryCommands::Recall { bank_id, query, fact_type, budget, max_tokens, trace, include_chunks, chunk_max_tokens } => {
                 commands::memory::recall(&client, &bank_id, query, fact_type, budget, max_tokens, trace, include_chunks, chunk_max_tokens, verbose, output_format)
             }
@@ -446,6 +764,7 @@ fn run() -> Result<()> {
             }
         },
 
+        // Document commands
         Commands::Document(doc_cmd) => match doc_cmd {
             DocumentCommands::List { bank_id, query, limit, offset } => {
                 commands::document::list(&client, &bank_id, query, limit, offset, verbose, output_format)
@@ -458,6 +777,7 @@ fn run() -> Result<()> {
             }
         },
 
+        // Entity commands
         Commands::Entity(entity_cmd) => match entity_cmd {
             EntityCommands::List { bank_id, limit } => {
                 commands::entity::list(&client, &bank_id, limit, verbose, output_format)
@@ -470,14 +790,73 @@ fn run() -> Result<()> {
             }
         },
 
+        // Tag commands
+        Commands::Tag(tag_cmd) => match tag_cmd {
+            TagCommands::List { bank_id, query, limit, offset } => {
+                commands::tag::list(&client, &bank_id, query, limit, offset, verbose, output_format)
+            }
+        },
+
+        // Chunk commands
+        Commands::Chunk(chunk_cmd) => match chunk_cmd {
+            ChunkCommands::Get { chunk_id } => {
+                commands::chunk::get(&client, &chunk_id, verbose, output_format)
+            }
+        },
+
+        // Operation commands
         Commands::Operation(op_cmd) => match op_cmd {
             OperationCommands::List { bank_id } => {
                 commands::operation::list(&client, &bank_id, verbose, output_format)
             }
+            OperationCommands::Get { bank_id, operation_id } => {
+                commands::operation::get(&client, &bank_id, &operation_id, verbose, output_format)
+            }
             OperationCommands::Cancel { bank_id, operation_id } => {
                 commands::operation::cancel(&client, &bank_id, &operation_id, verbose, output_format)
             }
         },
+
+        // Reflection commands
+        Commands::Reflection(ref_cmd) => match ref_cmd {
+            ReflectionCommands::List { bank_id } => {
+                commands::reflection::list(&client, &bank_id, verbose, output_format)
+            }
+            ReflectionCommands::Get { bank_id, reflection_id } => {
+                commands::reflection::get(&client, &bank_id, &reflection_id, verbose, output_format)
+            }
+            ReflectionCommands::Create { bank_id, name, source_query } => {
+                commands::reflection::create(&client, &bank_id, &name, &source_query, verbose, output_format)
+            }
+            ReflectionCommands::Update { bank_id, reflection_id, name } => {
+                commands::reflection::update(&client, &bank_id, &reflection_id, name, verbose, output_format)
+            }
+            ReflectionCommands::Delete { bank_id, reflection_id, yes } => {
+                commands::reflection::delete(&client, &bank_id, &reflection_id, yes, verbose, output_format)
+            }
+            ReflectionCommands::Refresh { bank_id, reflection_id } => {
+                commands::reflection::refresh(&client, &bank_id, &reflection_id, verbose, output_format)
+            }
+        },
+
+        // Directive commands
+        Commands::Directive(dir_cmd) => match dir_cmd {
+            DirectiveCommands::List { bank_id } => {
+                commands::directive::list(&client, &bank_id, verbose, output_format)
+            }
+            DirectiveCommands::Get { bank_id, directive_id } => {
+                commands::directive::get(&client, &bank_id, &directive_id, verbose, output_format)
+            }
+            DirectiveCommands::Create { bank_id, name, content } => {
+                commands::directive::create(&client, &bank_id, &name, &content, verbose, output_format)
+            }
+            DirectiveCommands::Update { bank_id, directive_id, name, content } => {
+                commands::directive::update(&client, &bank_id, &directive_id, name, content, verbose, output_format)
+            }
+            DirectiveCommands::Delete { bank_id, directive_id, yes } => {
+                commands::directive::delete(&client, &bank_id, &directive_id, yes, verbose, output_format)
+            }
+        },
     };
 
     // Handle API errors with nice messages

hindsight-cli/src/ui.rs

@@ -173,7 +173,7 @@ pub fn print_think_response(response: &ReflectResponse) {
     println!();
 
     if let Some(based_on) = &response.based_on {
-        let count = based_on.memories.len() + based_on.mental_models.len();
+        let count = based_on.memories.len();
         if count > 0 {
             println!("{}", dim(&format!("Based on {} memory units", count)));
         }

hindsight-cli/tests/cli_integration.rs

@@ -0,0 +1,483 @@
+//! Integration tests for the hindsight CLI commands.
+//!
+//! These tests require a running hindsight API server.
+//! Set HINDSIGHT_API_URL environment variable to point to the server.
+//! Tests will be skipped if the server is not available.
+
+use std::env;
+use std::process::Command;
+
+/// Check if the API server is available
+fn server_available() -> bool {
+    let api_url = env::var("HINDSIGHT_API_URL").unwrap_or_else(|_| "http://localhost:8080".to_string());
+    let health_url = format!("{}/health", api_url);
+
+    match reqwest::blocking::get(&health_url) {
+        Ok(resp) => resp.status().is_success(),
+        Err(_) => false,
+    }
+}
+
+/// Helper macro to skip tests when server is not available
+macro_rules! skip_if_no_server {
+    () => {
+        if !server_available() {
+            eprintln!("Skipping test: API server not available");
+            return;
+        }
+    };
+}
+
+/// Get the path to the hindsight binary
+fn hindsight_binary() -> String {
+    env::var("CARGO_BIN_EXE_hindsight")
+        .unwrap_or_else(|_| {
+            // Try common locations
+            let target_debug = "./target/debug/hindsight";
+            let target_release = "./target/release/hindsight";
+            if std::path::Path::new(target_debug).exists() {
+                target_debug.to_string()
+            } else if std::path::Path::new(target_release).exists() {
+                target_release.to_string()
+            } else {
+                "hindsight".to_string()
+            }
+        })
+}
+
+/// Test bank ID for integration tests - each test needs a unique bank ID
+/// to avoid parallel test interference
+fn test_bank_id(test_name: &str) -> String {
+    format!("cli-test-{}-{}", test_name, std::process::id())
+}
+
+/// Run a hindsight CLI command
+fn run_hindsight(args: &[&str]) -> std::process::Output {
+    let api_url = env::var("HINDSIGHT_API_URL").unwrap_or_else(|_| "http://localhost:8080".to_string());
+
+    Command::new(hindsight_binary())
+        .env("HINDSIGHT_API_URL", &api_url)
+        .args(args)
+        .output()
+        .expect("Failed to execute hindsight command")
+}
+
+#[test]
+fn test_health_check() {
+    skip_if_no_server!();
+
+    let output = run_hindsight(&["health"]);
+
+    // Should succeed or fail gracefully
+    let stdout = String::from_utf8_lossy(&output.stdout);
+    let stderr = String::from_utf8_lossy(&output.stderr);
+
+    // Either succeeded with "healthy" output or has a reasonable error
+    if output.status.success() {
+        // Note: output may contain ANSI color codes, so check for key text
+        assert!(
+            stdout.contains("healthy") || stdout.contains("Health") || stdout.contains("status"),
+            "Expected health check output, got: {} / {}",
+            stdout,
+            stderr
+        );
+    }
+}
+
+#[test]
+fn test_health_check_json_output() {
+    skip_if_no_server!();
+
+    let output = run_hindsight(&["health", "-o", "json"]);
+
+    if output.status.success() {
+        let stdout = String::from_utf8_lossy(&output.stdout);
+        // Should be valid JSON
+        let result: serde_json::Value = serde_json::from_str(&stdout)
+            .expect(&format!("Expected valid JSON output, got: {}", stdout));
+
+        // Should have status field
+        assert!(result.get("status").is_some(), "Expected status field in health response");
+    }
+}
+
+#[test]
+fn test_bank_list() {
+    skip_if_no_server!();
+
+    let output = run_hindsight(&["bank", "list"]);
+
+    // Should succeed (even if no banks exist)
+    let stdout = String::from_utf8_lossy(&output.stdout);
+    let stderr = String::from_utf8_lossy(&output.stderr);
+
+    assert!(
+        output.status.success(),
+        "Bank list command failed: {} / {}",
+        stdout,
+        stderr
+    );
+}
+
+#[test]
+fn test_bank_list_json_output() {
+    skip_if_no_server!();
+
+    let output = run_hindsight(&["bank", "list", "-o", "json"]);
+
+    if output.status.success() {
+        let stdout = String::from_utf8_lossy(&output.stdout);
+        // Should be valid JSON array
+        let _result: serde_json::Value = serde_json::from_str(&stdout)
+            .expect(&format!("Expected valid JSON output, got: {}", stdout));
+    }
+}
+
+#[test]
+fn test_bank_create_and_delete() {
+    skip_if_no_server!();
+
+    let bank_id = test_bank_id("create-delete");
+
+    // Create a bank
+    let output = run_hindsight(&[
+        "bank", "create",
+        &bank_id,
+        "--name", "Test Bank",
+        "--mission", "A test bank for CLI integration tests",
+    ]);
+
+    let stdout = String::from_utf8_lossy(&output.stdout);
+    let stderr = String::from_utf8_lossy(&output.stderr);
+
+    // Bank might already exist, which is OK
+    let created = output.status.success();
+
+    // Get bank disposition
+    let output = run_hindsight(&["bank", "disposition", &bank_id]);
+    if created {
+        assert!(
+            output.status.success(),
+            "Bank disposition command failed: {} / {}",
+            String::from_utf8_lossy(&output.stdout),
+            String::from_utf8_lossy(&output.stderr)
+        );
+    }
+
+    // Clean up: delete the bank
+    let output = run_hindsight(&["bank", "delete", &bank_id, "-y"]);
+    // Deletion should succeed
+    if created {
+        assert!(
+            output.status.success(),
+            "Bank delete command failed: {} / {}",
+            String::from_utf8_lossy(&output.stdout),
+            String::from_utf8_lossy(&output.stderr)
+        );
+    }
+}
+
+#[test]
+fn test_memory_list() {
+    skip_if_no_server!();
+
+    let bank_id = test_bank_id("memory-list");
+
+    // Create the bank first
+    let _ = run_hindsight(&["bank", "create", &bank_id, "--name", "Test Bank"]);
+
+    // List memories (should be empty for new bank)
+    let output = run_hindsight(&["memory", "list", &bank_id]);
+
+    let stdout = String::from_utf8_lossy(&output.stdout);
+    let stderr = String::from_utf8_lossy(&output.stderr);
+
+    // Should succeed (even if empty)
+    assert!(
+        output.status.success(),
+        "Memory list command failed: {} / {}",
+        stdout,
+        stderr
+    );
+
+    // Clean up
+    let _ = run_hindsight(&["bank", "delete", &bank_id, "-y"]);
+}
+
+#[test]
+fn test_mental_model_list() {
+    skip_if_no_server!();
+
+    let bank_id = test_bank_id("mm-list");
+
+    // Create the bank first
+    let _ = run_hindsight(&["bank", "create", &bank_id, "--name", "Test Bank"]);
+
+    // List mental models
+    let output = run_hindsight(&["mental-model", "list", &bank_id]);
+
+    let stdout = String::from_utf8_lossy(&output.stdout);
+    let stderr = String::from_utf8_lossy(&output.stderr);
+
+    // Should succeed
+    assert!(
+        output.status.success(),
+        "Mental model list command failed: {} / {}",
+        stdout,
+        stderr
+    );
+
+    // Clean up
+    let _ = run_hindsight(&["bank", "delete", &bank_id, "-y"]);
+}
+
+#[test]
+fn test_mental_model_create_and_delete() {
+    skip_if_no_server!();
+
+    let bank_id = test_bank_id("mm-create");
+
+    // Create the bank first
+    let _ = run_hindsight(&["bank", "create", &bank_id, "--name", "Test Bank"]);
+
+    // Create a mental model
+    let output = run_hindsight(&[
+        "mental-model", "create",
+        &bank_id,
+        "Test Model",
+        "A test mental model",
+    ]);
+
+    let stdout = String::from_utf8_lossy(&output.stdout);
+    let stderr = String::from_utf8_lossy(&output.stderr);
+
+    // The create command should succeed
+    assert!(
+        output.status.success(),
+        "Mental model create failed: stdout={}, stderr={}",
+        stdout,
+        stderr
+    );
+
+    // Verify it's in the list
+    let output = run_hindsight(&["mental-model", "list", &bank_id, "-o", "json"]);
+    let stdout = String::from_utf8_lossy(&output.stdout);
+
+    assert!(
+        output.status.success(),
+        "Mental model list failed: {}",
+        stdout
+    );
+
+    // Parse JSON and verify model exists
+    if let Ok(result) = serde_json::from_str::<serde_json::Value>(&stdout) {
+        if let Some(items) = result.get("items").and_then(|v| v.as_array()) {
+            // Check if any model has the name "Test Model"
+            let found = items.iter().any(|item| {
+                item.get("name").and_then(|v| v.as_str()) == Some("Test Model")
+            });
+            assert!(found, "Expected to find 'Test Model' in mental models list: {}", stdout);
+        }
+    }
+
+    // Clean up
+    let _ = run_hindsight(&["bank", "delete", &bank_id, "-y"]);
+}
+
+#[test]
+fn test_tag_list() {
+    skip_if_no_server!();
+
+    let bank_id = test_bank_id("tag-list");
+
+    // Create the bank first
+    let _ = run_hindsight(&["bank", "create", &bank_id, "--name", "Test Bank"]);
+
+    // List tags
+    let output = run_hindsight(&["tag", "list", &bank_id]);
+
+    let stdout = String::from_utf8_lossy(&output.stdout);
+    let stderr = String::from_utf8_lossy(&output.stderr);
+
+    // Should succeed (even if no tags)
+    assert!(
+        output.status.success(),
+        "Tag list command failed: {} / {}",
+        stdout,
+        stderr
+    );
+
+    // Clean up
+    let _ = run_hindsight(&["bank", "delete", &bank_id, "-y"]);
+}
+
+#[test]
+fn test_entity_list() {
+    skip_if_no_server!();
+
+    let bank_id = test_bank_id("entity-list");
+
+    // Create the bank first
+    let _ = run_hindsight(&["bank", "create", &bank_id, "--name", "Test Bank"]);
+
+    // List entities
+    let output = run_hindsight(&["entity", "list", &bank_id]);
+
+    let stdout = String::from_utf8_lossy(&output.stdout);
+    let stderr = String::from_utf8_lossy(&output.stderr);
+
+    // Should succeed (even if no entities)
+    assert!(
+        output.status.success(),
+        "Entity list command failed: {} / {}",
+        stdout,
+        stderr
+    );
+
+    // Clean up
+    let _ = run_hindsight(&["bank", "delete", &bank_id, "-y"]);
+}
+
+#[test]
+fn test_operation_list() {
+    skip_if_no_server!();
+
+    let bank_id = test_bank_id("op-list");
+
+    // Create the bank first
+    let _ = run_hindsight(&["bank", "create", &bank_id, "--name", "Test Bank"]);
+
+    // List operations
+    let output = run_hindsight(&["operation", "list", &bank_id]);
+
+    let stdout = String::from_utf8_lossy(&output.stdout);
+    let stderr = String::from_utf8_lossy(&output.stderr);
+
+    // Should succeed (even if no operations)
+    assert!(
+        output.status.success(),
+        "Operation list command failed: {} / {}",
+        stdout,
+        stderr
+    );
+
+    // Clean up
+    let _ = run_hindsight(&["bank", "delete", &bank_id, "-y"]);
+}
+
+#[test]
+fn test_bank_stats() {
+    skip_if_no_server!();
+
+    let bank_id = test_bank_id("bank-stats");
+
+    // Create the bank first
+    let _ = run_hindsight(&["bank", "create", &bank_id, "--name", "Test Bank"]);
+
+    // Get stats
+    let output = run_hindsight(&["bank", "stats", &bank_id]);
+
+    let stdout = String::from_utf8_lossy(&output.stdout);
+    let stderr = String::from_utf8_lossy(&output.stderr);
+
+    // Should succeed
+    assert!(
+        output.status.success(),
+        "Bank stats command failed: {} / {}",
+        stdout,
+        stderr
+    );
+
+    // Clean up
+    let _ = run_hindsight(&["bank", "delete", &bank_id, "-y"]);
+}
+
+#[test]
+fn test_bank_graph() {
+    skip_if_no_server!();
+
+    let bank_id = test_bank_id("bank-graph");
+
+    // Create the bank first
+    let _ = run_hindsight(&["bank", "create", &bank_id, "--name", "Test Bank"]);
+
+    // Get graph
+    let output = run_hindsight(&["bank", "graph", &bank_id]);
+
+    let stdout = String::from_utf8_lossy(&output.stdout);
+    let stderr = String::from_utf8_lossy(&output.stderr);
+
+    // Should succeed (even if empty graph)
+    assert!(
+        output.status.success(),
+        "Bank graph command failed: {} / {}",
+        stdout,
+        stderr
+    );
+
+    // Clean up
+    let _ = run_hindsight(&["bank", "delete", &bank_id, "-y"]);
+}
+
+#[test]
+fn test_bank_update() {
+    skip_if_no_server!();
+
+    let bank_id = test_bank_id("bank-update");
+
+    // Create the bank first
+    let output = run_hindsight(&["bank", "create", &bank_id, "--name", "Test Bank"]);
+
+    if output.status.success() {
+        // Update the bank
+        let output = run_hindsight(&[
+            "bank", "update", &bank_id,
+            "--name", "Updated Test Bank",
+        ]);
+
+        let stdout = String::from_utf8_lossy(&output.stdout);
+        let stderr = String::from_utf8_lossy(&output.stderr);
+
+        assert!(
+            output.status.success(),
+            "Bank update command failed: {} / {}",
+            stdout,
+            stderr
+        );
+
+        // Verify the update
+        let output = run_hindsight(&["bank", "disposition", &bank_id, "-o", "json"]);
+        if output.status.success() {
+            let stdout = String::from_utf8_lossy(&output.stdout);
+            let result: serde_json::Value = serde_json::from_str(&stdout).unwrap();
+            assert_eq!(
+                result.get("name").and_then(|v| v.as_str()),
+                Some("Updated Test Bank")
+            );
+        }
+    }
+
+    // Clean up
+    let _ = run_hindsight(&["bank", "delete", &bank_id, "-y"]);
+}
+
+#[test]
+fn test_json_yaml_output_formats() {
+    skip_if_no_server!();
+
+    // Test JSON output for bank list
+    let output = run_hindsight(&["bank", "list", "-o", "json"]);
+    if output.status.success() {
+        let stdout = String::from_utf8_lossy(&output.stdout);
+        let _: serde_json::Value = serde_json::from_str(&stdout)
+            .expect("Expected valid JSON for bank list");
+    }
+
+    // Test YAML output for bank list
+    let output = run_hindsight(&["bank", "list", "-o", "yaml"]);
+    if output.status.success() {
+        let stdout = String::from_utf8_lossy(&output.stdout);
+        let _: serde_yaml::Value = serde_yaml::from_str(&stdout)
+            .expect("Expected valid YAML for bank list");
+    }
+}

hindsight-clients/python/.openapi-generator/FILES

@@ -1,19 +1,19 @@
 hindsight_client_api/__init__.py
 hindsight_client_api/api/__init__.py
 hindsight_client_api/api/banks_api.py
+hindsight_client_api/api/directives_api.py
 hindsight_client_api/api/documents_api.py
 hindsight_client_api/api/entities_api.py
 hindsight_client_api/api/memory_api.py
-hindsight_client_api/api/mental_models_api.py
 hindsight_client_api/api/monitoring_api.py
 hindsight_client_api/api/operations_api.py
+hindsight_client_api/api/reflections_api.py
 hindsight_client_api/api_client.py
 hindsight_client_api/api_response.py
 hindsight_client_api/configuration.py
 hindsight_client_api/exceptions.py
 hindsight_client_api/models/__init__.py
 hindsight_client_api/models/add_background_request.py
-hindsight_client_api/models/async_operation_submit_response.py
 hindsight_client_api/models/background_response.py
 hindsight_client_api/models/bank_list_item.py
 hindsight_client_api/models/bank_list_response.py
@@ -24,11 +24,15 @@ hindsight_client_api/models/cancel_operation_response.py
 hindsight_client_api/models/chunk_data.py
 hindsight_client_api/models/chunk_include_options.py
 hindsight_client_api/models/chunk_response.py
+hindsight_client_api/models/consolidation_response.py
 hindsight_client_api/models/create_bank_request.py
-hindsight_client_api/models/create_mental_model_request.py
-hindsight_client_api/models/created_mental_model.py
+hindsight_client_api/models/create_directive_request.py
+hindsight_client_api/models/create_reflection_request.py
+hindsight_client_api/models/create_reflection_response.py
 hindsight_client_api/models/delete_document_response.py
 hindsight_client_api/models/delete_response.py
+hindsight_client_api/models/directive_list_response.py
+hindsight_client_api/models/directive_response.py
 hindsight_client_api/models/disposition_traits.py
 hindsight_client_api/models/document_response.py
 hindsight_client_api/models/entity_detail_response.py
@@ -38,6 +42,7 @@ hindsight_client_api/models/entity_list_item.py
 hindsight_client_api/models/entity_list_response.py
 hindsight_client_api/models/entity_observation_response.py
 hindsight_client_api/models/entity_state_response.py
+hindsight_client_api/models/features_info.py
 hindsight_client_api/models/graph_data_response.py
 hindsight_client_api/models/http_validation_error.py
 hindsight_client_api/models/include_options.py
@@ -45,12 +50,6 @@ hindsight_client_api/models/list_documents_response.py
 hindsight_client_api/models/list_memory_units_response.py
 hindsight_client_api/models/list_tags_response.py
 hindsight_client_api/models/memory_item.py
-hindsight_client_api/models/mental_model_freshness_response.py
-hindsight_client_api/models/mental_model_list_response.py
-hindsight_client_api/models/mental_model_observation_response.py
-hindsight_client_api/models/mental_model_response.py
-hindsight_client_api/models/observation_evidence_response.py
-hindsight_client_api/models/observation_input.py
 hindsight_client_api/models/operation_response.py
 hindsight_client_api/models/operation_status_response.py
 hindsight_client_api/models/operations_list_response.py
@@ -66,15 +65,18 @@ hindsight_client_api/models/reflect_request.py
 hindsight_client_api/models/reflect_response.py
 hindsight_client_api/models/reflect_tool_call.py
 hindsight_client_api/models/reflect_trace.py
-hindsight_client_api/models/refresh_mental_models_request.py
+hindsight_client_api/models/reflection_list_response.py
+hindsight_client_api/models/reflection_response.py
 hindsight_client_api/models/retain_request.py
 hindsight_client_api/models/retain_response.py
 hindsight_client_api/models/tag_item.py
 hindsight_client_api/models/token_usage.py
 hindsight_client_api/models/tool_calls_include_options.py
+hindsight_client_api/models/update_directive_request.py
 hindsight_client_api/models/update_disposition_request.py
-hindsight_client_api/models/update_mental_model_request.py
+hindsight_client_api/models/update_reflection_request.py
 hindsight_client_api/models/validation_error.py
 hindsight_client_api/models/validation_error_loc_inner.py
+hindsight_client_api/models/version_response.py
 hindsight_client_api/rest.py
 hindsight_client_api_README.md

hindsight-clients/python/hindsight_client/__init__.py

@@ -25,17 +25,18 @@ Example:
     ```
 """
 
-from .hindsight_client import Hindsight
+from hindsight_client_api.models.bank_profile_response import BankProfileResponse
+from hindsight_client_api.models.disposition_traits import DispositionTraits
+from hindsight_client_api.models.list_memory_units_response import ListMemoryUnitsResponse
+from hindsight_client_api.models.recall_response import RecallResponse as _RecallResponse
+from hindsight_client_api.models.recall_result import RecallResult as _RecallResult
+from hindsight_client_api.models.reflect_fact import ReflectFact
+from hindsight_client_api.models.reflect_response import ReflectResponse
 
 # Re-export response types for convenient access
 from hindsight_client_api.models.retain_response import RetainResponse
-from hindsight_client_api.models.recall_response import RecallResponse as _RecallResponse
-from hindsight_client_api.models.recall_result import RecallResult as _RecallResult
-from hindsight_client_api.models.reflect_response import ReflectResponse
-from hindsight_client_api.models.reflect_fact import ReflectFact
-from hindsight_client_api.models.list_memory_units_response import ListMemoryUnitsResponse
-from hindsight_client_api.models.bank_profile_response import BankProfileResponse
-from hindsight_client_api.models.disposition_traits import DispositionTraits
+
+from .hindsight_client import Hindsight
 
 
 # Add cleaner __repr__ and __iter__ for REPL usability

hindsight-clients/python/hindsight_client/hindsight_client.py

@@ -6,23 +6,23 @@ easy-to-use interface on top of the auto-generated OpenAPI client.
 """
 
 import asyncio
-from typing import Optional, List, Dict, Any
 from datetime import datetime
+from typing import Any, Literal
 
 import hindsight_client_api
-from hindsight_client_api.api import memory_api, banks_api
+from hindsight_client_api.api import banks_api, memory_api
 from hindsight_client_api.models import (
-    recall_request,
-    retain_request,
     memory_item,
+    recall_request,
     reflect_request,
+    retain_request,
 )
-from hindsight_client_api.models.retain_response import RetainResponse
+from hindsight_client_api.models.bank_profile_response import BankProfileResponse
+from hindsight_client_api.models.list_memory_units_response import ListMemoryUnitsResponse
 from hindsight_client_api.models.recall_response import RecallResponse
 from hindsight_client_api.models.recall_result import RecallResult
 from hindsight_client_api.models.reflect_response import ReflectResponse
-from hindsight_client_api.models.list_memory_units_response import ListMemoryUnitsResponse
-from hindsight_client_api.models.bank_profile_response import BankProfileResponse
+from hindsight_client_api.models.retain_response import RetainResponse
 
 
 def _run_async(coro):
@@ -63,7 +63,7 @@ class Hindsight:
         ```
     """
 
-    def __init__(self, base_url: str, api_key: Optional[str] = None, timeout: float = 30.0):
+    def __init__(self, base_url: str, api_key: str | None = None, timeout: float = 30.0):
         """
         Initialize the Hindsight client.
 
@@ -110,12 +110,12 @@ class Hindsight:
         self,
         bank_id: str,
         content: str,
-        timestamp: Optional[datetime] = None,
-        context: Optional[str] = None,
-        document_id: Optional[str] = None,
-        metadata: Optional[Dict[str, str]] = None,
-        entities: Optional[List[Dict[str, str]]] = None,
-        tags: Optional[List[str]] = None,
+        timestamp: datetime | None = None,
+        context: str | None = None,
+        document_id: str | None = None,
+        metadata: dict[str, str] | None = None,
+        entities: list[dict[str, str]] | None = None,
+        tags: list[str] | None = None,
     ) -> RetainResponse:
         """
         Store a single memory (simplified interface).
@@ -128,24 +128,33 @@ class Hindsight:
             document_id: Optional document ID for grouping
             metadata: Optional user-defined metadata
             entities: Optional list of entities [{"text": "...", "type": "..."}]
-            tags: Optional list of tags for this memory
+            tags: Optional list of tags for filtering memories during recall/reflect
 
         Returns:
             RetainResponse with success status
         """
         return self.retain_batch(
             bank_id=bank_id,
-            items=[{"content": content, "timestamp": timestamp, "context": context, "metadata": metadata, "entities": entities, "tags": tags}],
+            items=[
+                {
+                    "content": content,
+                    "timestamp": timestamp,
+                    "context": context,
+                    "metadata": metadata,
+                    "entities": entities,
+                    "tags": tags,
+                }
+            ],
             document_id=document_id,
         )
 
     def retain_batch(
         self,
         bank_id: str,
-        items: List[Dict[str, Any]],
-        document_id: Optional[str] = None,
+        items: list[dict[str, Any]],
+        document_id: str | None = None,
+        document_tags: list[str] | None = None,
         retain_async: bool = False,
-        document_tags: Optional[List[str]] = None,
     ) -> RetainResponse:
         """
         Store multiple memories in batch.
@@ -154,8 +163,8 @@ class Hindsight:
             bank_id: The memory bank ID
             items: List of memory items with 'content' and optional 'timestamp', 'context', 'metadata', 'document_id', 'entities', 'tags'
             document_id: Optional document ID for grouping memories (applied to items that don't have their own)
+            document_tags: Optional list of tags applied to all items in this batch (merged with per-item tags)
             retain_async: If True, process asynchronously in background (default: False)
-            document_tags: Optional list of tags to apply to all memories in this batch
 
         Returns:
             RetainResponse with success status and item count
@@ -166,10 +175,7 @@ class Hindsight:
         for item in items:
             entities = None
             if item.get("entities"):
-                entities = [
-                    EntityInput(text=e["text"], type=e.get("type"))
-                    for e in item["entities"]
-                ]
+                entities = [EntityInput(text=e["text"], type=e.get("type")) for e in item["entities"]]
             memory_items.append(
                 memory_item.MemoryItem(
                     content=item["content"],
@@ -195,17 +201,17 @@ class Hindsight:
         self,
         bank_id: str,
         query: str,
-        types: Optional[List[str]] = None,
+        types: list[str] | None = None,
         max_tokens: int = 4096,
         budget: str = "mid",
         trace: bool = False,
-        query_timestamp: Optional[str] = None,
+        query_timestamp: str | None = None,
         include_entities: bool = False,
         max_entity_tokens: int = 500,
         include_chunks: bool = False,
         max_chunk_tokens: int = 8192,
-        tags: Optional[List[str]] = None,
-        tags_match: str = "any",
+        tags: list[str] | None = None,
+        tags_match: Literal["any", "all", "any_strict", "all_strict"] = "any",
     ) -> RecallResponse:
         """
         Recall memories using semantic similarity.
@@ -223,16 +229,18 @@ class Hindsight:
             include_chunks: Include raw text chunks in results (default: False)
             max_chunk_tokens: Maximum tokens for chunks (default: 8192)
             tags: Optional list of tags to filter memories by
-            tags_match: How to match tags: 'any' (OR, includes untagged), 'all' (AND, includes untagged),
-                'any_strict' (OR, excludes untagged), 'all_strict' (AND, excludes untagged). Default: 'any'
+            tags_match: How to match tags - "any" (OR, includes untagged), "all" (AND, includes untagged),
+                "any_strict" (OR, excludes untagged), "all_strict" (AND, excludes untagged). Default: "any"
 
         Returns:
             RecallResponse with results, optional entities, optional chunks, and optional trace
         """
-        from hindsight_client_api.models import include_options, entity_include_options, chunk_include_options
+        from hindsight_client_api.models import chunk_include_options, entity_include_options, include_options
 
         include_opts = include_options.IncludeOptions(
-            entities=entity_include_options.EntityIncludeOptions(max_tokens=max_entity_tokens) if include_entities else None,
+            entities=entity_include_options.EntityIncludeOptions(max_tokens=max_entity_tokens)
+            if include_entities
+            else None,
             chunks=chunk_include_options.ChunkIncludeOptions(max_tokens=max_chunk_tokens) if include_chunks else None,
         )
 
@@ -255,11 +263,11 @@ class Hindsight:
         bank_id: str,
         query: str,
         budget: str = "low",
-        context: Optional[str] = None,
-        max_tokens: Optional[int] = None,
-        response_schema: Optional[Dict[str, Any]] = None,
-        tags: Optional[List[str]] = None,
-        tags_match: str = "any",
+        context: str | None = None,
+        max_tokens: int | None = None,
+        response_schema: dict[str, Any] | None = None,
+        tags: list[str] | None = None,
+        tags_match: Literal["any", "all", "any_strict", "all_strict"] = "any",
     ) -> ReflectResponse:
         """
         Generate a contextual answer based on bank identity and memories.
@@ -274,8 +282,8 @@ class Hindsight:
                 the response will include a 'structured_output' field with the LLM
                 response parsed according to this schema.
             tags: Optional list of tags to filter memories by
-            tags_match: How to match tags: 'any' (OR, includes untagged), 'all' (AND, includes untagged),
-                'any_strict' (OR, excludes untagged), 'all_strict' (AND, excludes untagged). Default: 'any'
+            tags_match: How to match tags - "any" (OR, includes untagged), "all" (AND, includes untagged),
+                "any_strict" (OR, excludes untagged), "all_strict" (AND, excludes untagged). Default: "any"
 
         Returns:
             ReflectResponse with answer text, optionally facts used, and optionally
@@ -296,26 +304,28 @@ class Hindsight:
     def list_memories(
         self,
         bank_id: str,
-        type: Optional[str] = None,
-        search_query: Optional[str] = None,
+        type: str | None = None,
+        search_query: str | None = None,
         limit: int = 100,
         offset: int = 0,
     ) -> ListMemoryUnitsResponse:
         """List memory units with pagination."""
-        return _run_async(self._memory_api.list_memories(
-            bank_id=bank_id,
-            type=type,
-            q=search_query,
-            limit=limit,
-            offset=offset,
-        ))
+        return _run_async(
+            self._memory_api.list_memories(
+                bank_id=bank_id,
+                type=type,
+                q=search_query,
+                limit=limit,
+                offset=offset,
+            )
+        )
 
     def create_bank(
         self,
         bank_id: str,
-        name: Optional[str] = None,
-        background: Optional[str] = None,
-        disposition: Optional[Dict[str, float]] = None,
+        name: str | None = None,
+        background: str | None = None,
+        disposition: dict[str, float] | None = None,
     ) -> BankProfileResponse:
         """Create or update a memory bank."""
         from hindsight_client_api.models import create_bank_request, disposition_traits
@@ -332,13 +342,34 @@ class Hindsight:
 
         return _run_async(self._banks_api.create_or_update_bank(bank_id, request_obj))
 
+    def set_mission(
+        self,
+        bank_id: str,
+        mission: str,
+    ) -> BankProfileResponse:
+        """
+        Set or update the mission for a memory bank.
+
+        Args:
+            bank_id: The memory bank ID
+            mission: The mission text describing the agent's purpose
+
+        Returns:
+            BankProfileResponse with updated bank profile
+        """
+        from hindsight_client_api.models import create_bank_request
+
+        request_obj = create_bank_request.CreateBankRequest(mission=mission)
+        return _run_async(self._banks_api.create_or_update_bank(bank_id, request_obj))
+
     # Async methods (native async, no _run_async wrapper)
 
     async def aretain_batch(
         self,
         bank_id: str,
-        items: List[Dict[str, Any]],
-        document_id: Optional[str] = None,
+        items: list[dict[str, Any]],
+        document_id: str | None = None,
+        document_tags: list[str] | None = None,
         retain_async: bool = False,
     ) -> RetainResponse:
         """
@@ -346,8 +377,9 @@ class Hindsight:
 
         Args:
             bank_id: The memory bank ID
-            items: List of memory items with 'content' and optional 'timestamp', 'context', 'metadata', 'document_id', 'entities'
+            items: List of memory items with 'content' and optional 'timestamp', 'context', 'metadata', 'document_id', 'entities', 'tags'
             document_id: Optional document ID for grouping memories (applied to items that don't have their own)
+            document_tags: Optional list of tags applied to all items in this batch (merged with per-item tags)
             retain_async: If True, process asynchronously in background (default: False)
 
         Returns:
@@ -359,10 +391,7 @@ class Hindsight:
         for item in items:
             entities = None
             if item.get("entities"):
-                entities = [
-                    EntityInput(text=e["text"], type=e.get("type"))
-                    for e in item["entities"]
-                ]
+                entities = [EntityInput(text=e["text"], type=e.get("type")) for e in item["entities"]]
             memory_items.append(
                 memory_item.MemoryItem(
                     content=item["content"],
@@ -372,12 +401,14 @@ class Hindsight:
                     # Use item's document_id if provided, otherwise fall back to batch-level document_id
                     document_id=item.get("document_id") or document_id,
                     entities=entities,
+                    tags=item.get("tags"),
                 )
             )
 
         request_obj = retain_request.RetainRequest(
             items=memory_items,
             async_=retain_async,
+            document_tags=document_tags,
         )
 
         return await self._memory_api.retain_memories(bank_id, request_obj)
@@ -386,11 +417,12 @@ class Hindsight:
         self,
         bank_id: str,
         content: str,
-        timestamp: Optional[datetime] = None,
-        context: Optional[str] = None,
-        document_id: Optional[str] = None,
-        metadata: Optional[Dict[str, str]] = None,
-        entities: Optional[List[Dict[str, str]]] = None,
+        timestamp: datetime | None = None,
+        context: str | None = None,
+        document_id: str | None = None,
+        metadata: dict[str, str] | None = None,
+        entities: list[dict[str, str]] | None = None,
+        tags: list[str] | None = None,
     ) -> RetainResponse:
         """
         Store a single memory (async).
@@ -403,13 +435,23 @@ class Hindsight:
             document_id: Optional document ID for grouping
             metadata: Optional user-defined metadata
             entities: Optional list of entities [{"text": "...", "type": "..."}]
+            tags: Optional list of tags for filtering memories during recall/reflect
 
         Returns:
             RetainResponse with success status
         """
         return await self.aretain_batch(
             bank_id=bank_id,
-            items=[{"content": content, "timestamp": timestamp, "context": context, "metadata": metadata, "entities": entities}],
+            items=[
+                {
+                    "content": content,
+                    "timestamp": timestamp,
+                    "context": context,
+                    "metadata": metadata,
+                    "entities": entities,
+                    "tags": tags,
+                }
+            ],
             document_id=document_id,
         )
 
@@ -417,10 +459,12 @@ class Hindsight:
         self,
         bank_id: str,
         query: str,
-        types: Optional[List[str]] = None,
+        types: list[str] | None = None,
         max_tokens: int = 4096,
         budget: str = "mid",
-    ) -> List[RecallResult]:
+        tags: list[str] | None = None,
+        tags_match: Literal["any", "all", "any_strict", "all_strict"] = "any",
+    ) -> list[RecallResult]:
         """
         Recall memories using semantic similarity (async).
 
@@ -430,6 +474,9 @@ class Hindsight:
             types: Optional list of fact types to filter (world, experience, opinion, observation)
             max_tokens: Maximum tokens in results (default: 4096)
             budget: Budget level for recall - "low", "mid", or "high" (default: "mid")
+            tags: Optional list of tags to filter memories by
+            tags_match: How to match tags - "any" (OR, includes untagged), "all" (AND, includes untagged),
+                "any_strict" (OR, excludes untagged), "all_strict" (AND, excludes untagged). Default: "any"
 
         Returns:
             List of RecallResult objects
@@ -440,17 +487,21 @@ class Hindsight:
             budget=budget,
             max_tokens=max_tokens,
             trace=False,
+            tags=tags,
+            tags_match=tags_match,
         )
 
         response = await self._memory_api.recall_memories(bank_id, request_obj)
-        return response.results if hasattr(response, 'results') else []
+        return response.results if hasattr(response, "results") else []
 
     async def areflect(
         self,
         bank_id: str,
         query: str,
         budget: str = "low",
-        context: Optional[str] = None,
+        context: str | None = None,
+        tags: list[str] | None = None,
+        tags_match: Literal["any", "all", "any_strict", "all_strict"] = "any",
     ) -> ReflectResponse:
         """
         Generate a contextual answer based on bank identity and memories (async).
@@ -460,6 +511,9 @@ class Hindsight:
             query: The question or prompt
             budget: Budget level for reflection - "low", "mid", or "high" (default: "low")
             context: Optional additional context
+            tags: Optional list of tags to filter memories by
+            tags_match: How to match tags - "any" (OR, includes untagged), "all" (AND, includes untagged),
+                "any_strict" (OR, excludes untagged), "all_strict" (AND, excludes untagged). Default: "any"
 
         Returns:
             ReflectResponse with answer text and optionally facts used
@@ -468,6 +522,8 @@ class Hindsight:
             query=query,
             budget=budget,
             context=context,
+            tags=tags,
+            tags_match=tags_match,
         )
 
         return await self._memory_api.reflect(bank_id, request_obj)

hindsight-clients/python/hindsight_client_api/__init__.py

@@ -18,12 +18,13 @@ __version__ = "0.0.7"
 
 # import apis into sdk package
 from hindsight_client_api.api.banks_api import BanksApi
+from hindsight_client_api.api.directives_api import DirectivesApi
 from hindsight_client_api.api.documents_api import DocumentsApi
 from hindsight_client_api.api.entities_api import EntitiesApi
 from hindsight_client_api.api.memory_api import MemoryApi
-from hindsight_client_api.api.mental_models_api import MentalModelsApi
 from hindsight_client_api.api.monitoring_api import MonitoringApi
 from hindsight_client_api.api.operations_api import OperationsApi
+from hindsight_client_api.api.reflections_api import ReflectionsApi
 
 # import ApiClient
 from hindsight_client_api.api_response import ApiResponse
@@ -38,7 +39,6 @@ from hindsight_client_api.exceptions import ApiException
 
 # import models into sdk package
 from hindsight_client_api.models.add_background_request import AddBackgroundRequest
-from hindsight_client_api.models.async_operation_submit_response import AsyncOperationSubmitResponse
 from hindsight_client_api.models.background_response import BackgroundResponse
 from hindsight_client_api.models.bank_list_item import BankListItem
 from hindsight_client_api.models.bank_list_response import BankListResponse
@@ -49,11 +49,15 @@ from hindsight_client_api.models.cancel_operation_response import CancelOperatio
 from hindsight_client_api.models.chunk_data import ChunkData
 from hindsight_client_api.models.chunk_include_options import ChunkIncludeOptions
 from hindsight_client_api.models.chunk_response import ChunkResponse
+from hindsight_client_api.models.consolidation_response import ConsolidationResponse
 from hindsight_client_api.models.create_bank_request import CreateBankRequest
-from hindsight_client_api.models.create_mental_model_request import CreateMentalModelRequest
-from hindsight_client_api.models.created_mental_model import CreatedMentalModel
+from hindsight_client_api.models.create_directive_request import CreateDirectiveRequest
+from hindsight_client_api.models.create_reflection_request import CreateReflectionRequest
+from hindsight_client_api.models.create_reflection_response import CreateReflectionResponse
 from hindsight_client_api.models.delete_document_response import DeleteDocumentResponse
 from hindsight_client_api.models.delete_response import DeleteResponse
+from hindsight_client_api.models.directive_list_response import DirectiveListResponse
+from hindsight_client_api.models.directive_response import DirectiveResponse
 from hindsight_client_api.models.disposition_traits import DispositionTraits
 from hindsight_client_api.models.document_response import DocumentResponse
 from hindsight_client_api.models.entity_detail_response import EntityDetailResponse
@@ -63,6 +67,7 @@ from hindsight_client_api.models.entity_list_item import EntityListItem
 from hindsight_client_api.models.entity_list_response import EntityListResponse
 from hindsight_client_api.models.entity_observation_response import EntityObservationResponse
 from hindsight_client_api.models.entity_state_response import EntityStateResponse
+from hindsight_client_api.models.features_info import FeaturesInfo
 from hindsight_client_api.models.graph_data_response import GraphDataResponse
 from hindsight_client_api.models.http_validation_error import HTTPValidationError
 from hindsight_client_api.models.include_options import IncludeOptions
@@ -70,12 +75,6 @@ from hindsight_client_api.models.list_documents_response import ListDocumentsRes
 from hindsight_client_api.models.list_memory_units_response import ListMemoryUnitsResponse
 from hindsight_client_api.models.list_tags_response import ListTagsResponse
 from hindsight_client_api.models.memory_item import MemoryItem
-from hindsight_client_api.models.mental_model_freshness_response import MentalModelFreshnessResponse
-from hindsight_client_api.models.mental_model_list_response import MentalModelListResponse
-from hindsight_client_api.models.mental_model_observation_response import MentalModelObservationResponse
-from hindsight_client_api.models.mental_model_response import MentalModelResponse
-from hindsight_client_api.models.observation_evidence_response import ObservationEvidenceResponse
-from hindsight_client_api.models.observation_input import ObservationInput
 from hindsight_client_api.models.operation_response import OperationResponse
 from hindsight_client_api.models.operation_status_response import OperationStatusResponse
 from hindsight_client_api.models.operations_list_response import OperationsListResponse
@@ -91,13 +90,16 @@ from hindsight_client_api.models.reflect_request import ReflectRequest
 from hindsight_client_api.models.reflect_response import ReflectResponse
 from hindsight_client_api.models.reflect_tool_call import ReflectToolCall
 from hindsight_client_api.models.reflect_trace import ReflectTrace
-from hindsight_client_api.models.refresh_mental_models_request import RefreshMentalModelsRequest
+from hindsight_client_api.models.reflection_list_response import ReflectionListResponse
+from hindsight_client_api.models.reflection_response import ReflectionResponse
 from hindsight_client_api.models.retain_request import RetainRequest
 from hindsight_client_api.models.retain_response import RetainResponse
 from hindsight_client_api.models.tag_item import TagItem
 from hindsight_client_api.models.token_usage import TokenUsage
 from hindsight_client_api.models.tool_calls_include_options import ToolCallsIncludeOptions
+from hindsight_client_api.models.update_directive_request import UpdateDirectiveRequest
 from hindsight_client_api.models.update_disposition_request import UpdateDispositionRequest
-from hindsight_client_api.models.update_mental_model_request import UpdateMentalModelRequest
+from hindsight_client_api.models.update_reflection_request import UpdateReflectionRequest
 from hindsight_client_api.models.validation_error import ValidationError
 from hindsight_client_api.models.validation_error_loc_inner import ValidationErrorLocInner
+from hindsight_client_api.models.version_response import VersionResponse

hindsight-clients/python/hindsight_client_api/api/__init__.py

@@ -2,10 +2,11 @@
 
 # import apis into api package
 from hindsight_client_api.api.banks_api import BanksApi
+from hindsight_client_api.api.directives_api import DirectivesApi
 from hindsight_client_api.api.documents_api import DocumentsApi
 from hindsight_client_api.api.entities_api import EntitiesApi
 from hindsight_client_api.api.memory_api import MemoryApi
-from hindsight_client_api.api.mental_models_api import MentalModelsApi
 from hindsight_client_api.api.monitoring_api import MonitoringApi
 from hindsight_client_api.api.operations_api import OperationsApi
+from hindsight_client_api.api.reflections_api import ReflectionsApi
 

hindsight-clients/python/hindsight_client_api/api/banks_api.py

@@ -23,6 +23,7 @@ from hindsight_client_api.models.background_response import BackgroundResponse
 from hindsight_client_api.models.bank_list_response import BankListResponse
 from hindsight_client_api.models.bank_profile_response import BankProfileResponse
 from hindsight_client_api.models.bank_stats_response import BankStatsResponse
+from hindsight_client_api.models.consolidation_response import ConsolidationResponse
 from hindsight_client_api.models.create_bank_request import CreateBankRequest
 from hindsight_client_api.models.delete_response import DeleteResponse
 from hindsight_client_api.models.update_disposition_request import UpdateDispositionRequest
@@ -354,6 +355,284 @@ class BanksApi:
 
 
 
+    @validate_call
+    async def clear_mental_models(
+        self,
+        bank_id: StrictStr,
+        authorization: Optional[StrictStr] = None,
+        _request_timeout: Union[
+            None,
+            Annotated[StrictFloat, Field(gt=0)],
+            Tuple[
+                Annotated[StrictFloat, Field(gt=0)],
+                Annotated[StrictFloat, Field(gt=0)]
+            ]
+        ] = None,
+        _request_auth: Optional[Dict[StrictStr, Any]] = None,
+        _content_type: Optional[StrictStr] = None,
+        _headers: Optional[Dict[StrictStr, Any]] = None,
+        _host_index: Annotated[StrictInt, Field(ge=0, le=0)] = 0,
+    ) -> DeleteResponse:
+        """Clear all mental models
+
+        Delete all mental models for a memory bank. This is useful for resetting the consolidated knowledge.
+
+        :param bank_id: (required)
+        :type bank_id: str
+        :param authorization:
+        :type authorization: str
+        :param _request_timeout: timeout setting for this request. If one
+                                 number provided, it will be total request
+                                 timeout. It can also be a pair (tuple) of
+                                 (connection, read) timeouts.
+        :type _request_timeout: int, tuple(int, int), optional
+        :param _request_auth: set to override the auth_settings for an a single
+                              request; this effectively ignores the
+                              authentication in the spec for a single request.
+        :type _request_auth: dict, optional
+        :param _content_type: force content-type for the request.
+        :type _content_type: str, Optional
+        :param _headers: set to override the headers for a single
+                         request; this effectively ignores the headers
+                         in the spec for a single request.
+        :type _headers: dict, optional
+        :param _host_index: set to override the host_index for a single
+                            request; this effectively ignores the host_index
+                            in the spec for a single request.
+        :type _host_index: int, optional
+        :return: Returns the result object.
+        """ # noqa: E501
+
+        _param = self._clear_mental_models_serialize(
+            bank_id=bank_id,
+            authorization=authorization,
+            _request_auth=_request_auth,
+            _content_type=_content_type,
+            _headers=_headers,
+            _host_index=_host_index
+        )
+
+        _response_types_map: Dict[str, Optional[str]] = {
+            '200': "DeleteResponse",
+            '422': "HTTPValidationError",
+        }
+        response_data = await self.api_client.call_api(
+            *_param,
+            _request_timeout=_request_timeout
+        )
+        await response_data.read()
+        return self.api_client.response_deserialize(
+            response_data=response_data,
+            response_types_map=_response_types_map,
+        ).data
+
+
+    @validate_call
+    async def clear_mental_models_with_http_info(
+        self,
+        bank_id: StrictStr,
+        authorization: Optional[StrictStr] = None,
+        _request_timeout: Union[
+            None,
+            Annotated[StrictFloat, Field(gt=0)],
+            Tuple[
+                Annotated[StrictFloat, Field(gt=0)],
+                Annotated[StrictFloat, Field(gt=0)]
+            ]
+        ] = None,
+        _request_auth: Optional[Dict[StrictStr, Any]] = None,
+        _content_type: Optional[StrictStr] = None,
+        _headers: Optional[Dict[StrictStr, Any]] = None,
+        _host_index: Annotated[StrictInt, Field(ge=0, le=0)] = 0,
+    ) -> ApiResponse[DeleteResponse]:
+        """Clear all mental models
+
+        Delete all mental models for a memory bank. This is useful for resetting the consolidated knowledge.
+
+        :param bank_id: (required)
+        :type bank_id: str
+        :param authorization:
+        :type authorization: str
+        :param _request_timeout: timeout setting for this request. If one
+                                 number provided, it will be total request
+                                 timeout. It can also be a pair (tuple) of
+                                 (connection, read) timeouts.
+        :type _request_timeout: int, tuple(int, int), optional
+        :param _request_auth: set to override the auth_settings for an a single
+                              request; this effectively ignores the
+                              authentication in the spec for a single request.
+        :type _request_auth: dict, optional
+        :param _content_type: force content-type for the request.
+        :type _content_type: str, Optional
+        :param _headers: set to override the headers for a single
+                         request; this effectively ignores the headers
+                         in the spec for a single request.
+        :type _headers: dict, optional
+        :param _host_index: set to override the host_index for a single
+                            request; this effectively ignores the host_index
+                            in the spec for a single request.
+        :type _host_index: int, optional
+        :return: Returns the result object.
+        """ # noqa: E501
+
+        _param = self._clear_mental_models_serialize(
+            bank_id=bank_id,
+            authorization=authorization,
+            _request_auth=_request_auth,
+            _content_type=_content_type,
+            _headers=_headers,
+            _host_index=_host_index
+        )
+
+        _response_types_map: Dict[str, Optional[str]] = {
+            '200': "DeleteResponse",
+            '422': "HTTPValidationError",
+        }
+        response_data = await self.api_client.call_api(
+            *_param,
+            _request_timeout=_request_timeout
+        )
+        await response_data.read()
+        return self.api_client.response_deserialize(
+            response_data=response_data,
+            response_types_map=_response_types_map,
+        )
+
+
+    @validate_call
+    async def clear_mental_models_without_preload_content(
+        self,
+        bank_id: StrictStr,
+        authorization: Optional[StrictStr] = None,
+        _request_timeout: Union[
+            None,
+            Annotated[StrictFloat, Field(gt=0)],
+            Tuple[
+                Annotated[StrictFloat, Field(gt=0)],
+                Annotated[StrictFloat, Field(gt=0)]
+            ]
+        ] = None,
+        _request_auth: Optional[Dict[StrictStr, Any]] = None,
+        _content_type: Optional[StrictStr] = None,
+        _headers: Optional[Dict[StrictStr, Any]] = None,
+        _host_index: Annotated[StrictInt, Field(ge=0, le=0)] = 0,
+    ) -> RESTResponseType:
+        """Clear all mental models
+
+        Delete all mental models for a memory bank. This is useful for resetting the consolidated knowledge.
+
+        :param bank_id: (required)
+        :type bank_id: str
+        :param authorization:
+        :type authorization: str
+        :param _request_timeout: timeout setting for this request. If one
+                                 number provided, it will be total request
+                                 timeout. It can also be a pair (tuple) of
+                                 (connection, read) timeouts.
+        :type _request_timeout: int, tuple(int, int), optional
+        :param _request_auth: set to override the auth_settings for an a single
+                              request; this effectively ignores the
+                              authentication in the spec for a single request.
+        :type _request_auth: dict, optional
+        :param _content_type: force content-type for the request.
+        :type _content_type: str, Optional
+        :param _headers: set to override the headers for a single
+                         request; this effectively ignores the headers
+                         in the spec for a single request.
+        :type _headers: dict, optional
+        :param _host_index: set to override the host_index for a single
+                            request; this effectively ignores the host_index
+                            in the spec for a single request.
+        :type _host_index: int, optional
+        :return: Returns the result object.
+        """ # noqa: E501
+
+        _param = self._clear_mental_models_serialize(
+            bank_id=bank_id,
+            authorization=authorization,
+            _request_auth=_request_auth,
+            _content_type=_content_type,
+            _headers=_headers,
+            _host_index=_host_index
+        )
+
+        _response_types_map: Dict[str, Optional[str]] = {
+            '200': "DeleteResponse",
+            '422': "HTTPValidationError",
+        }
+        response_data = await self.api_client.call_api(
+            *_param,
+            _request_timeout=_request_timeout
+        )
+        return response_data.response
+
+
+    def _clear_mental_models_serialize(
+        self,
+        bank_id,
+        authorization,
+        _request_auth,
+        _content_type,
+        _headers,
+        _host_index,
+    ) -> RequestSerialized:
+
+        _host = None
+
+        _collection_formats: Dict[str, str] = {
+        }
+
+        _path_params: Dict[str, str] = {}
+        _query_params: List[Tuple[str, str]] = []
+        _header_params: Dict[str, Optional[str]] = _headers or {}
+        _form_params: List[Tuple[str, str]] = []
+        _files: Dict[
+            str, Union[str, bytes, List[str], List[bytes], List[Tuple[str, bytes]]]
+        ] = {}
+        _body_params: Optional[bytes] = None
+
+        # process the path parameters
+        if bank_id is not None:
+            _path_params['bank_id'] = bank_id
+        # process the query parameters
+        # process the header parameters
+        if authorization is not None:
+            _header_params['authorization'] = authorization
+        # process the form parameters
+        # process the body parameter
+
+
+        # set the HTTP header `Accept`
+        if 'Accept' not in _header_params:
+            _header_params['Accept'] = self.api_client.select_header_accept(
+                [
+                    'application/json'
+                ]
+            )
+
+
+        # authentication setting
+        _auth_settings: List[str] = [
+        ]
+
+        return self.api_client.param_serialize(
+            method='DELETE',
+            resource_path='/v1/default/banks/{bank_id}/mental-models',
+            path_params=_path_params,
+            query_params=_query_params,
+            header_params=_header_params,
+            body=_body_params,
+            post_params=_form_params,
+            files=_files,
+            auth_settings=_auth_settings,
+            collection_formats=_collection_formats,
+            _host=_host,
+            _request_auth=_request_auth
+        )
+
+
+
+
     @validate_call
     async def create_or_update_bank(
         self,
@@ -1757,6 +2036,284 @@ class BanksApi:
 
 
 
+    @validate_call
+    async def trigger_consolidation(
+        self,
+        bank_id: StrictStr,
+        authorization: Optional[StrictStr] = None,
+        _request_timeout: Union[
+            None,
+            Annotated[StrictFloat, Field(gt=0)],
+            Tuple[
+                Annotated[StrictFloat, Field(gt=0)],
+                Annotated[StrictFloat, Field(gt=0)]
+            ]
+        ] = None,
+        _request_auth: Optional[Dict[StrictStr, Any]] = None,
+        _content_type: Optional[StrictStr] = None,
+        _headers: Optional[Dict[StrictStr, Any]] = None,
+        _host_index: Annotated[StrictInt, Field(ge=0, le=0)] = 0,
+    ) -> ConsolidationResponse:
+        """Trigger consolidation
+
+        Run memory consolidation to create/update mental models from recent memories.
+
+        :param bank_id: (required)
+        :type bank_id: str
+        :param authorization:
+        :type authorization: str
+        :param _request_timeout: timeout setting for this request. If one
+                                 number provided, it will be total request
+                                 timeout. It can also be a pair (tuple) of
+                                 (connection, read) timeouts.
+        :type _request_timeout: int, tuple(int, int), optional
+        :param _request_auth: set to override the auth_settings for an a single
+                              request; this effectively ignores the
+                              authentication in the spec for a single request.
+        :type _request_auth: dict, optional
+        :param _content_type: force content-type for the request.
+        :type _content_type: str, Optional
+        :param _headers: set to override the headers for a single
+                         request; this effectively ignores the headers
+                         in the spec for a single request.
+        :type _headers: dict, optional
+        :param _host_index: set to override the host_index for a single
+                            request; this effectively ignores the host_index
+                            in the spec for a single request.
+        :type _host_index: int, optional
+        :return: Returns the result object.
+        """ # noqa: E501
+
+        _param = self._trigger_consolidation_serialize(
+            bank_id=bank_id,
+            authorization=authorization,
+            _request_auth=_request_auth,
+            _content_type=_content_type,
+            _headers=_headers,
+            _host_index=_host_index
+        )
+
+        _response_types_map: Dict[str, Optional[str]] = {
+            '200': "ConsolidationResponse",
+            '422': "HTTPValidationError",
+        }
+        response_data = await self.api_client.call_api(
+            *_param,
+            _request_timeout=_request_timeout
+        )
+        await response_data.read()
+        return self.api_client.response_deserialize(
+            response_data=response_data,
+            response_types_map=_response_types_map,
+        ).data
+
+
+    @validate_call
+    async def trigger_consolidation_with_http_info(
+        self,
+        bank_id: StrictStr,
+        authorization: Optional[StrictStr] = None,
+        _request_timeout: Union[
+            None,
+            Annotated[StrictFloat, Field(gt=0)],
+            Tuple[
+                Annotated[StrictFloat, Field(gt=0)],
+                Annotated[StrictFloat, Field(gt=0)]
+            ]
+        ] = None,
+        _request_auth: Optional[Dict[StrictStr, Any]] = None,
+        _content_type: Optional[StrictStr] = None,
+        _headers: Optional[Dict[StrictStr, Any]] = None,
+        _host_index: Annotated[StrictInt, Field(ge=0, le=0)] = 0,
+    ) -> ApiResponse[ConsolidationResponse]:
+        """Trigger consolidation
+
+        Run memory consolidation to create/update mental models from recent memories.
+
+        :param bank_id: (required)
+        :type bank_id: str
+        :param authorization:
+        :type authorization: str
+        :param _request_timeout: timeout setting for this request. If one
+                                 number provided, it will be total request
+                                 timeout. It can also be a pair (tuple) of
+                                 (connection, read) timeouts.
+        :type _request_timeout: int, tuple(int, int), optional
+        :param _request_auth: set to override the auth_settings for an a single
+                              request; this effectively ignores the
+                              authentication in the spec for a single request.
+        :type _request_auth: dict, optional
+        :param _content_type: force content-type for the request.
+        :type _content_type: str, Optional
+        :param _headers: set to override the headers for a single
+                         request; this effectively ignores the headers
+                         in the spec for a single request.
+        :type _headers: dict, optional
+        :param _host_index: set to override the host_index for a single
+                            request; this effectively ignores the host_index
+                            in the spec for a single request.
+        :type _host_index: int, optional
+        :return: Returns the result object.
+        """ # noqa: E501
+
+        _param = self._trigger_consolidation_serialize(
+            bank_id=bank_id,
+            authorization=authorization,
+            _request_auth=_request_auth,
+            _content_type=_content_type,
+            _headers=_headers,
+            _host_index=_host_index
+        )
+
+        _response_types_map: Dict[str, Optional[str]] = {
+            '200': "ConsolidationResponse",
+            '422': "HTTPValidationError",
+        }
+        response_data = await self.api_client.call_api(
+            *_param,
+            _request_timeout=_request_timeout
+        )
+        await response_data.read()
+        return self.api_client.response_deserialize(
+            response_data=response_data,
+            response_types_map=_response_types_map,
+        )
+
+
+    @validate_call
+    async def trigger_consolidation_without_preload_content(
+        self,
+        bank_id: StrictStr,
+        authorization: Optional[StrictStr] = None,
+        _request_timeout: Union[
+            None,
+            Annotated[StrictFloat, Field(gt=0)],
+            Tuple[
+                Annotated[StrictFloat, Field(gt=0)],
+                Annotated[StrictFloat, Field(gt=0)]
+            ]
+        ] = None,
+        _request_auth: Optional[Dict[StrictStr, Any]] = None,
+        _content_type: Optional[StrictStr] = None,
+        _headers: Optional[Dict[StrictStr, Any]] = None,
+        _host_index: Annotated[StrictInt, Field(ge=0, le=0)] = 0,
+    ) -> RESTResponseType:
+        """Trigger consolidation
+
+        Run memory consolidation to create/update mental models from recent memories.
+
+        :param bank_id: (required)
+        :type bank_id: str
+        :param authorization:
+        :type authorization: str
+        :param _request_timeout: timeout setting for this request. If one
+                                 number provided, it will be total request
+                                 timeout. It can also be a pair (tuple) of
+                                 (connection, read) timeouts.
+        :type _request_timeout: int, tuple(int, int), optional
+        :param _request_auth: set to override the auth_settings for an a single
+                              request; this effectively ignores the
+                              authentication in the spec for a single request.
+        :type _request_auth: dict, optional
+        :param _content_type: force content-type for the request.
+        :type _content_type: str, Optional
+        :param _headers: set to override the headers for a single
+                         request; this effectively ignores the headers
+                         in the spec for a single request.
+        :type _headers: dict, optional
+        :param _host_index: set to override the host_index for a single
+                            request; this effectively ignores the host_index
+                            in the spec for a single request.
+        :type _host_index: int, optional
+        :return: Returns the result object.
+        """ # noqa: E501
+
+        _param = self._trigger_consolidation_serialize(
+            bank_id=bank_id,
+            authorization=authorization,
+            _request_auth=_request_auth,
+            _content_type=_content_type,
+            _headers=_headers,
+            _host_index=_host_index
+        )
+
+        _response_types_map: Dict[str, Optional[str]] = {
+            '200': "ConsolidationResponse",
+            '422': "HTTPValidationError",
+        }
+        response_data = await self.api_client.call_api(
+            *_param,
+            _request_timeout=_request_timeout
+        )
+        return response_data.response
+
+
+    def _trigger_consolidation_serialize(
+        self,
+        bank_id,
+        authorization,
+        _request_auth,
+        _content_type,
+        _headers,
+        _host_index,
+    ) -> RequestSerialized:
+
+        _host = None
+
+        _collection_formats: Dict[str, str] = {
+        }
+
+        _path_params: Dict[str, str] = {}
+        _query_params: List[Tuple[str, str]] = []
+        _header_params: Dict[str, Optional[str]] = _headers or {}
+        _form_params: List[Tuple[str, str]] = []
+        _files: Dict[
+            str, Union[str, bytes, List[str], List[bytes], List[Tuple[str, bytes]]]
+        ] = {}
+        _body_params: Optional[bytes] = None
+
+        # process the path parameters
+        if bank_id is not None:
+            _path_params['bank_id'] = bank_id
+        # process the query parameters
+        # process the header parameters
+        if authorization is not None:
+            _header_params['authorization'] = authorization
+        # process the form parameters
+        # process the body parameter
+
+
+        # set the HTTP header `Accept`
+        if 'Accept' not in _header_params:
+            _header_params['Accept'] = self.api_client.select_header_accept(
+                [
+                    'application/json'
+                ]
+            )
+
+
+        # authentication setting
+        _auth_settings: List[str] = [
+        ]
+
+        return self.api_client.param_serialize(
+            method='POST',
+            resource_path='/v1/default/banks/{bank_id}/consolidate',
+            path_params=_path_params,
+            query_params=_query_params,
+            header_params=_header_params,
+            body=_body_params,
+            post_params=_form_params,
+            files=_files,
+            auth_settings=_auth_settings,
+            collection_formats=_collection_formats,
+            _host=_host,
+            _request_auth=_request_auth
+        )
+
+
+
+
     @validate_call
     async def update_bank(
         self,

hindsight-clients/python/hindsight_client_api/api/monitoring_api.py

@@ -17,6 +17,7 @@ from typing import Any, Dict, List, Optional, Tuple, Union
 from typing_extensions import Annotated
 
 from typing import Any
+from hindsight_client_api.models.version_response import VersionResponse
 
 from hindsight_client_api.api_client import ApiClient, RequestSerialized
 from hindsight_client_api.api_response import ApiResponse
@@ -36,6 +37,251 @@ class MonitoringApi:
         self.api_client = api_client
 
 
+    @validate_call
+    async def get_version(
+        self,
+        _request_timeout: Union[
+            None,
+            Annotated[StrictFloat, Field(gt=0)],
+            Tuple[
+                Annotated[StrictFloat, Field(gt=0)],
+                Annotated[StrictFloat, Field(gt=0)]
+            ]
+        ] = None,
+        _request_auth: Optional[Dict[StrictStr, Any]] = None,
+        _content_type: Optional[StrictStr] = None,
+        _headers: Optional[Dict[StrictStr, Any]] = None,
+        _host_index: Annotated[StrictInt, Field(ge=0, le=0)] = 0,
+    ) -> VersionResponse:
+        """Get API version and feature flags
+
+        Returns API version information and enabled feature flags. Use this to check which capabilities are available in this deployment.
+
+        :param _request_timeout: timeout setting for this request. If one
+                                 number provided, it will be total request
+                                 timeout. It can also be a pair (tuple) of
+                                 (connection, read) timeouts.
+        :type _request_timeout: int, tuple(int, int), optional
+        :param _request_auth: set to override the auth_settings for an a single
+                              request; this effectively ignores the
+                              authentication in the spec for a single request.
+        :type _request_auth: dict, optional
+        :param _content_type: force content-type for the request.
+        :type _content_type: str, Optional
+        :param _headers: set to override the headers for a single
+                         request; this effectively ignores the headers
+                         in the spec for a single request.
+        :type _headers: dict, optional
+        :param _host_index: set to override the host_index for a single
+                            request; this effectively ignores the host_index
+                            in the spec for a single request.
+        :type _host_index: int, optional
+        :return: Returns the result object.
+        """ # noqa: E501
+
+        _param = self._get_version_serialize(
+            _request_auth=_request_auth,
+            _content_type=_content_type,
+            _headers=_headers,
+            _host_index=_host_index
+        )
+
+        _response_types_map: Dict[str, Optional[str]] = {
+            '200': "VersionResponse",
+        }
+        response_data = await self.api_client.call_api(
+            *_param,
+            _request_timeout=_request_timeout
+        )
+        await response_data.read()
+        return self.api_client.response_deserialize(
+            response_data=response_data,
+            response_types_map=_response_types_map,
+        ).data
+
+
+    @validate_call
+    async def get_version_with_http_info(
+        self,
+        _request_timeout: Union[
+            None,
+            Annotated[StrictFloat, Field(gt=0)],
+            Tuple[
+                Annotated[StrictFloat, Field(gt=0)],
+                Annotated[StrictFloat, Field(gt=0)]
+            ]
+        ] = None,
+        _request_auth: Optional[Dict[StrictStr, Any]] = None,
+        _content_type: Optional[StrictStr] = None,
+        _headers: Optional[Dict[StrictStr, Any]] = None,
+        _host_index: Annotated[StrictInt, Field(ge=0, le=0)] = 0,
+    ) -> ApiResponse[VersionResponse]:
+        """Get API version and feature flags
+
+        Returns API version information and enabled feature flags. Use this to check which capabilities are available in this deployment.
+
+        :param _request_timeout: timeout setting for this request. If one
+                                 number provided, it will be total request
+                                 timeout. It can also be a pair (tuple) of
+                                 (connection, read) timeouts.
+        :type _request_timeout: int, tuple(int, int), optional
+        :param _request_auth: set to override the auth_settings for an a single
+                              request; this effectively ignores the
+                              authentication in the spec for a single request.
+        :type _request_auth: dict, optional
+        :param _content_type: force content-type for the request.
+        :type _content_type: str, Optional
+        :param _headers: set to override the headers for a single
+                         request; this effectively ignores the headers
+                         in the spec for a single request.
+        :type _headers: dict, optional
+        :param _host_index: set to override the host_index for a single
+                            request; this effectively ignores the host_index
+                            in the spec for a single request.
+        :type _host_index: int, optional
+        :return: Returns the result object.
+        """ # noqa: E501
+
+        _param = self._get_version_serialize(
+            _request_auth=_request_auth,
+            _content_type=_content_type,
+            _headers=_headers,
+            _host_index=_host_index
+        )
+
+        _response_types_map: Dict[str, Optional[str]] = {
+            '200': "VersionResponse",
+        }
+        response_data = await self.api_client.call_api(
+            *_param,
+            _request_timeout=_request_timeout
+        )
+        await response_data.read()
+        return self.api_client.response_deserialize(
+            response_data=response_data,
+            response_types_map=_response_types_map,
+        )
+
+
+    @validate_call
+    async def get_version_without_preload_content(
+        self,
+        _request_timeout: Union[
+            None,
+            Annotated[StrictFloat, Field(gt=0)],
+            Tuple[
+                Annotated[StrictFloat, Field(gt=0)],
+                Annotated[StrictFloat, Field(gt=0)]
+            ]
+        ] = None,
+        _request_auth: Optional[Dict[StrictStr, Any]] = None,
+        _content_type: Optional[StrictStr] = None,
+        _headers: Optional[Dict[StrictStr, Any]] = None,
+        _host_index: Annotated[StrictInt, Field(ge=0, le=0)] = 0,
+    ) -> RESTResponseType:
+        """Get API version and feature flags
+
+        Returns API version information and enabled feature flags. Use this to check which capabilities are available in this deployment.
+
+        :param _request_timeout: timeout setting for this request. If one
+                                 number provided, it will be total request
+                                 timeout. It can also be a pair (tuple) of
+                                 (connection, read) timeouts.
+        :type _request_timeout: int, tuple(int, int), optional
+        :param _request_auth: set to override the auth_settings for an a single
+                              request; this effectively ignores the
+                              authentication in the spec for a single request.
+        :type _request_auth: dict, optional
+        :param _content_type: force content-type for the request.
+        :type _content_type: str, Optional
+        :param _headers: set to override the headers for a single
+                         request; this effectively ignores the headers
+                         in the spec for a single request.
+        :type _headers: dict, optional
+        :param _host_index: set to override the host_index for a single
+                            request; this effectively ignores the host_index
+                            in the spec for a single request.
+        :type _host_index: int, optional
+        :return: Returns the result object.
+        """ # noqa: E501
+
+        _param = self._get_version_serialize(
+            _request_auth=_request_auth,
+            _content_type=_content_type,
+            _headers=_headers,
+            _host_index=_host_index
+        )
+
+        _response_types_map: Dict[str, Optional[str]] = {
+            '200': "VersionResponse",
+        }
+        response_data = await self.api_client.call_api(
+            *_param,
+            _request_timeout=_request_timeout
+        )
+        return response_data.response
+
+
+    def _get_version_serialize(
+        self,
+        _request_auth,
+        _content_type,
+        _headers,
+        _host_index,
+    ) -> RequestSerialized:
+
+        _host = None
+
+        _collection_formats: Dict[str, str] = {
+        }
+
+        _path_params: Dict[str, str] = {}
+        _query_params: List[Tuple[str, str]] = []
+        _header_params: Dict[str, Optional[str]] = _headers or {}
+        _form_params: List[Tuple[str, str]] = []
+        _files: Dict[
+            str, Union[str, bytes, List[str], List[bytes], List[Tuple[str, bytes]]]
+        ] = {}
+        _body_params: Optional[bytes] = None
+
+        # process the path parameters
+        # process the query parameters
+        # process the header parameters
+        # process the form parameters
+        # process the body parameter
+
+
+        # set the HTTP header `Accept`
+        if 'Accept' not in _header_params:
+            _header_params['Accept'] = self.api_client.select_header_accept(
+                [
+                    'application/json'
+                ]
+            )
+
+
+        # authentication setting
+        _auth_settings: List[str] = [
+        ]
+
+        return self.api_client.param_serialize(
+            method='GET',
+            resource_path='/version',
+            path_params=_path_params,
+            query_params=_query_params,
+            header_params=_header_params,
+            body=_body_params,
+            post_params=_form_params,
+            files=_files,
+            auth_settings=_auth_settings,
+            collection_formats=_collection_formats,
+            _host=_host,
+            _request_auth=_request_auth
+        )
+
+
+
+
     @validate_call
     async def health_endpoint_health_get(
         self,

hindsight-clients/python/hindsight_client_api/api/operations_api.py

@@ -16,8 +16,9 @@ from pydantic import validate_call, Field, StrictFloat, StrictStr, StrictInt
 from typing import Any, Dict, List, Optional, Tuple, Union
 from typing_extensions import Annotated
 
-from pydantic import StrictStr
+from pydantic import Field, StrictStr
 from typing import Optional
+from typing_extensions import Annotated
 from hindsight_client_api.models.cancel_operation_response import CancelOperationResponse
 from hindsight_client_api.models.operation_status_response import OperationStatusResponse
 from hindsight_client_api.models.operations_list_response import OperationsListResponse
@@ -630,6 +631,9 @@ class OperationsApi:
     async def list_operations(
         self,
         bank_id: StrictStr,
+        status: Annotated[Optional[StrictStr], Field(description="Filter by status: pending, completed, or failed")] = None,
+        limit: Annotated[Optional[Annotated[int, Field(le=100, strict=True, ge=1)]], Field(description="Maximum number of operations to return")] = None,
+        offset: Annotated[Optional[Annotated[int, Field(strict=True, ge=0)]], Field(description="Number of operations to skip")] = None,
         authorization: Optional[StrictStr] = None,
         _request_timeout: Union[
             None,
@@ -646,10 +650,16 @@ class OperationsApi:
     ) -> OperationsListResponse:
         """List async operations
 
-        Get a list of all async operations (pending and failed) for a specific agent, including error messages for failed operations
+        Get a list of async operations for a specific agent, with optional filtering by status. Results are sorted by most recent first.
 
         :param bank_id: (required)
         :type bank_id: str
+        :param status: Filter by status: pending, completed, or failed
+        :type status: str
+        :param limit: Maximum number of operations to return
+        :type limit: int
+        :param offset: Number of operations to skip
+        :type offset: int
         :param authorization:
         :type authorization: str
         :param _request_timeout: timeout setting for this request. If one
@@ -676,6 +686,9 @@ class OperationsApi:
 
         _param = self._list_operations_serialize(
             bank_id=bank_id,
+            status=status,
+            limit=limit,
+            offset=offset,
             authorization=authorization,
             _request_auth=_request_auth,
             _content_type=_content_type,
@@ -702,6 +715,9 @@ class OperationsApi:
     async def list_operations_with_http_info(
         self,
         bank_id: StrictStr,
+        status: Annotated[Optional[StrictStr], Field(description="Filter by status: pending, completed, or failed")] = None,
+        limit: Annotated[Optional[Annotated[int, Field(le=100, strict=True, ge=1)]], Field(description="Maximum number of operations to return")] = None,
+        offset: Annotated[Optional[Annotated[int, Field(strict=True, ge=0)]], Field(description="Number of operations to skip")] = None,
         authorization: Optional[StrictStr] = None,
         _request_timeout: Union[
             None,
@@ -718,10 +734,16 @@ class OperationsApi:
     ) -> ApiResponse[OperationsListResponse]:
         """List async operations
 
-        Get a list of all async operations (pending and failed) for a specific agent, including error messages for failed operations
+        Get a list of async operations for a specific agent, with optional filtering by status. Results are sorted by most recent first.
 
         :param bank_id: (required)
         :type bank_id: str
+        :param status: Filter by status: pending, completed, or failed
+        :type status: str
+        :param limit: Maximum number of operations to return
+        :type limit: int
+        :param offset: Number of operations to skip
+        :type offset: int
         :param authorization:
         :type authorization: str
         :param _request_timeout: timeout setting for this request. If one
@@ -748,6 +770,9 @@ class OperationsApi:
 
         _param = self._list_operations_serialize(
             bank_id=bank_id,
+            status=status,
+            limit=limit,
+            offset=offset,
             authorization=authorization,
             _request_auth=_request_auth,
             _content_type=_content_type,
@@ -774,6 +799,9 @@ class OperationsApi:
     async def list_operations_without_preload_content(
         self,
         bank_id: StrictStr,
+        status: Annotated[Optional[StrictStr], Field(description="Filter by status: pending, completed, or failed")] = None,
+        limit: Annotated[Optional[Annotated[int, Field(le=100, strict=True, ge=1)]], Field(description="Maximum number of operations to return")] = None,
+        offset: Annotated[Optional[Annotated[int, Field(strict=True, ge=0)]], Field(description="Number of operations to skip")] = None,
         authorization: Optional[StrictStr] = None,
         _request_timeout: Union[
             None,
@@ -790,10 +818,16 @@ class OperationsApi:
     ) -> RESTResponseType:
         """List async operations
 
-        Get a list of all async operations (pending and failed) for a specific agent, including error messages for failed operations
+        Get a list of async operations for a specific agent, with optional filtering by status. Results are sorted by most recent first.
 
         :param bank_id: (required)
         :type bank_id: str
+        :param status: Filter by status: pending, completed, or failed
+        :type status: str
+        :param limit: Maximum number of operations to return
+        :type limit: int
+        :param offset: Number of operations to skip
+        :type offset: int
         :param authorization:
         :type authorization: str
         :param _request_timeout: timeout setting for this request. If one
@@ -820,6 +854,9 @@ class OperationsApi:
 
         _param = self._list_operations_serialize(
             bank_id=bank_id,
+            status=status,
+            limit=limit,
+            offset=offset,
             authorization=authorization,
             _request_auth=_request_auth,
             _content_type=_content_type,
@@ -841,6 +878,9 @@ class OperationsApi:
     def _list_operations_serialize(
         self,
         bank_id,
+        status,
+        limit,
+        offset,
         authorization,
         _request_auth,
         _content_type,
@@ -866,6 +906,18 @@ class OperationsApi:
         if bank_id is not None:
             _path_params['bank_id'] = bank_id
         # process the query parameters
+        if status is not None:
+            
+            _query_params.append(('status', status))
+            
+        if limit is not None:
+            
+            _query_params.append(('limit', limit))
+            
+        if offset is not None:
+            
+            _query_params.append(('offset', offset))
+            
         # process the header parameters
         if authorization is not None:
             _header_params['authorization'] = authorization

hindsight-clients/python/hindsight_client_api/models/__init__.py

@@ -15,7 +15,6 @@
 
 # import models into model package
 from hindsight_client_api.models.add_background_request import AddBackgroundRequest
-from hindsight_client_api.models.async_operation_submit_response import AsyncOperationSubmitResponse
 from hindsight_client_api.models.background_response import BackgroundResponse
 from hindsight_client_api.models.bank_list_item import BankListItem
 from hindsight_client_api.models.bank_list_response import BankListResponse
@@ -26,11 +25,15 @@ from hindsight_client_api.models.cancel_operation_response import CancelOperatio
 from hindsight_client_api.models.chunk_data import ChunkData
 from hindsight_client_api.models.chunk_include_options import ChunkIncludeOptions
 from hindsight_client_api.models.chunk_response import ChunkResponse
+from hindsight_client_api.models.consolidation_response import ConsolidationResponse
 from hindsight_client_api.models.create_bank_request import CreateBankRequest
-from hindsight_client_api.models.create_mental_model_request import CreateMentalModelRequest
-from hindsight_client_api.models.created_mental_model import CreatedMentalModel
+from hindsight_client_api.models.create_directive_request import CreateDirectiveRequest
+from hindsight_client_api.models.create_reflection_request import CreateReflectionRequest
+from hindsight_client_api.models.create_reflection_response import CreateReflectionResponse
 from hindsight_client_api.models.delete_document_response import DeleteDocumentResponse
 from hindsight_client_api.models.delete_response import DeleteResponse
+from hindsight_client_api.models.directive_list_response import DirectiveListResponse
+from hindsight_client_api.models.directive_response import DirectiveResponse
 from hindsight_client_api.models.disposition_traits import DispositionTraits
 from hindsight_client_api.models.document_response import DocumentResponse
 from hindsight_client_api.models.entity_detail_response import EntityDetailResponse
@@ -40,6 +43,7 @@ from hindsight_client_api.models.entity_list_item import EntityListItem
 from hindsight_client_api.models.entity_list_response import EntityListResponse
 from hindsight_client_api.models.entity_observation_response import EntityObservationResponse
 from hindsight_client_api.models.entity_state_response import EntityStateResponse
+from hindsight_client_api.models.features_info import FeaturesInfo
 from hindsight_client_api.models.graph_data_response import GraphDataResponse
 from hindsight_client_api.models.http_validation_error import HTTPValidationError
 from hindsight_client_api.models.include_options import IncludeOptions
@@ -47,12 +51,6 @@ from hindsight_client_api.models.list_documents_response import ListDocumentsRes
 from hindsight_client_api.models.list_memory_units_response import ListMemoryUnitsResponse
 from hindsight_client_api.models.list_tags_response import ListTagsResponse
 from hindsight_client_api.models.memory_item import MemoryItem
-from hindsight_client_api.models.mental_model_freshness_response import MentalModelFreshnessResponse
-from hindsight_client_api.models.mental_model_list_response import MentalModelListResponse
-from hindsight_client_api.models.mental_model_observation_response import MentalModelObservationResponse
-from hindsight_client_api.models.mental_model_response import MentalModelResponse
-from hindsight_client_api.models.observation_evidence_response import ObservationEvidenceResponse
-from hindsight_client_api.models.observation_input import ObservationInput
 from hindsight_client_api.models.operation_response import OperationResponse
 from hindsight_client_api.models.operation_status_response import OperationStatusResponse
 from hindsight_client_api.models.operations_list_response import OperationsListResponse
@@ -68,13 +66,16 @@ from hindsight_client_api.models.reflect_request import ReflectRequest
 from hindsight_client_api.models.reflect_response import ReflectResponse
 from hindsight_client_api.models.reflect_tool_call import ReflectToolCall
 from hindsight_client_api.models.reflect_trace import ReflectTrace
-from hindsight_client_api.models.refresh_mental_models_request import RefreshMentalModelsRequest
+from hindsight_client_api.models.reflection_list_response import ReflectionListResponse
+from hindsight_client_api.models.reflection_response import ReflectionResponse
 from hindsight_client_api.models.retain_request import RetainRequest
 from hindsight_client_api.models.retain_response import RetainResponse
 from hindsight_client_api.models.tag_item import TagItem
 from hindsight_client_api.models.token_usage import TokenUsage
 from hindsight_client_api.models.tool_calls_include_options import ToolCallsIncludeOptions
+from hindsight_client_api.models.update_directive_request import UpdateDirectiveRequest
 from hindsight_client_api.models.update_disposition_request import UpdateDispositionRequest
-from hindsight_client_api.models.update_mental_model_request import UpdateMentalModelRequest
+from hindsight_client_api.models.update_reflection_request import UpdateReflectionRequest
 from hindsight_client_api.models.validation_error import ValidationError
 from hindsight_client_api.models.validation_error_loc_inner import ValidationErrorLocInner
+from hindsight_client_api.models.version_response import VersionResponse

hindsight-clients/python/hindsight_client_api/models/bank_stats_response.py

@@ -17,8 +17,8 @@ import pprint
 import re  # noqa: F401
 import json
 
-from pydantic import BaseModel, ConfigDict, StrictInt, StrictStr
-from typing import Any, ClassVar, Dict, List
+from pydantic import BaseModel, ConfigDict, Field, StrictInt, StrictStr
+from typing import Any, ClassVar, Dict, List, Optional
 from typing import Optional, Set
 from typing_extensions import Self
 
@@ -36,7 +36,10 @@ class BankStatsResponse(BaseModel):
     links_breakdown: Dict[str, Dict[str, StrictInt]]
     pending_operations: StrictInt
     failed_operations: StrictInt
-    __properties: ClassVar[List[str]] = ["bank_id", "total_nodes", "total_links", "total_documents", "nodes_by_fact_type", "links_by_link_type", "links_by_fact_type", "links_breakdown", "pending_operations", "failed_operations"]
+    last_consolidated_at: Optional[StrictStr] = None
+    pending_consolidation: Optional[StrictInt] = Field(default=0, description="Number of memories not yet processed into mental models")
+    total_mental_models: Optional[StrictInt] = Field(default=0, description="Total number of mental models")
+    __properties: ClassVar[List[str]] = ["bank_id", "total_nodes", "total_links", "total_documents", "nodes_by_fact_type", "links_by_link_type", "links_by_fact_type", "links_breakdown", "pending_operations", "failed_operations", "last_consolidated_at", "pending_consolidation", "total_mental_models"]
 
     model_config = ConfigDict(
         populate_by_name=True,
@@ -77,6 +80,11 @@ class BankStatsResponse(BaseModel):
             exclude=excluded_fields,
             exclude_none=True,
         )
+        # set to None if last_consolidated_at (nullable) is None
+        # and model_fields_set contains the field
+        if self.last_consolidated_at is None and "last_consolidated_at" in self.model_fields_set:
+            _dict['last_consolidated_at'] = None
+
         return _dict
 
     @classmethod
@@ -98,7 +106,10 @@ class BankStatsResponse(BaseModel):
             "links_by_fact_type": obj.get("links_by_fact_type"),
             "links_breakdown": obj.get("links_breakdown"),
             "pending_operations": obj.get("pending_operations"),
-            "failed_operations": obj.get("failed_operations")
+            "failed_operations": obj.get("failed_operations"),
+            "last_consolidated_at": obj.get("last_consolidated_at"),
+            "pending_consolidation": obj.get("pending_consolidation") if obj.get("pending_consolidation") is not None else 0,
+            "total_mental_models": obj.get("total_mental_models") if obj.get("total_mental_models") is not None else 0
         })
         return _obj