Save a fact to long-term memory (L4 CoreMemory).

Args: layer: "user" for user facts, "agent" for agent identity user_id: User identifier key: Fact key (e.g. "name", "language", "principle") value: Fact value importance: Importance score 0.0-1.0 (default 0.5) session_id: Session ID for dedup (optional)

Search memory across L3 (episodes) and L4 (facts).

Args: layer: "user" or "agent" user_id: User identifier query: Search query limit: Max results (default 10)

Delete a fact from L4 memory.

Args: layer: "user" or "agent" user_id: User identifier key: Fact key to delete

Start a new memory session.

Args: layer: "user" or "agent" user_id: User identifier

End a session and save summary.

Args: layer: "user" or "agent" user_id: User identifier session_id: Session ID from session_start summary: Session summary

Save an episode to L3 episodic memory.

Args: layer: "user" or "agent" user_id: User identifier summary: Episode description weight: Emotional weight 0.0-1.0 (default 0.5) tags: Tags (e.g. ["greeting", "decision", "error"])

Recall episodes, optionally filtered by tag.

Args: layer: "user" or "agent" user_id: User identifier tag: Filter by tag (empty = all) limit: Max results (default 10)

Add a node to the epistemic graph.

Args: layer: "user" or "agent" user_id: User identifier content: Node content node_type: Node type (fact, decision, emotion, error_analysis, etc.) tags: Tags

Query the epistemic graph by tag or node type.

Args: layer: "user" or "agent" user_id: User identifier tag: Filter by tag node_type: Filter by type limit: Max results

List recent memory sessions.

Args: layer: "user" or "agent" user_id: User identifier limit: Max results (default 10)

List episodes from L3 episodic memory.

Args: layer: "user" or "agent" user_id: User identifier limit: Max results (default 10) offset: Pagination offset

Get a single episode by ID.

Args: layer: "user" or "agent" user_id: User identifier episode_id: Episode database ID

List nodes from the epistemic graph.

Args: layer: "user" or "agent" user_id: User identifier node_type: Filter by type (empty = all types) limit: Max results (default 20)

List edges from the epistemic graph.

Args: layer: "user" or "agent" node_id: Source node ID (0 = all edges for the layer) limit: Max results (default 20)

Get memory statistics for a layer.

Args: layer: "user" or "agent" user_id: User identifier

Return compressed context summary for prompt injection (L4 top-10 + L3 top-3 + recent + wiki).

Args: layer: "user" or "agent" user_id: User identifier

Return compressed summary for prompt injection (L4 top-10 + L3 top-3).

Args: layer: "user" or "agent" user_id: User identifier

Manage API keys.

Args: action: "create", "revoke", or "list" user_id: User to create key for (create only) label: Optional label (create only) api_key: Key to revoke (revoke only)

Manage backups.

Args: action: "now", "list", "restore", or "status" backup_name: Backup to restore (restore only)

Run sagas with auto-rollback on failure.

Args: action: "consolidate" (gather->distill->promote) or "backup" (copy->verify) user_id: User identifier (consolidate only)

Import/export memory data.

Args: action: "export", "import", or "list" user_id: User to export (export only) file_path: File to import (import only) target_user_id: Import as this user (import only, defaults to user_id)

Sync read-only replica for dashboard/metrics.

Full memory cleanup: deduplicate, archive, clean staging.

Args: user_id: User identifier retention_days: Days to keep audit logs (default 30)

Emergency purge: delete all data from the last N hours.

Args: user_id: User identifier hours: How many hours back to purge (default 24)

Hybrid search across RAG + Wiki with strategy selection.

Args: query: Search query user_id: User identifier limit: Max results (default 10) strategy: "fts" (keyword), "mib" (semantic), "hybrid" (combined), or "auto" (auto-select) sources: "all" (RAG + Wiki), "rag" (RAG only), or "wiki" (Wiki only)