clude-memory-mcp

name	Clude Memory MCP
description	MCP server for Clude's cognitive memory system — store, recall, search, and dream. Supports hosted (zero-setup), self-hosted (Supabase + pgvector), and local (offline) modes.
metadata	{"openclaw":{"requires":{"bins":"[Truncated]"}}}

Clude Memory MCP

MCP server exposing a cognitive memory architecture inspired by Stanford's Generative Agents (Park et al. 2023).

Tools

recall_memories

Search the memory system. Returns scored memories ranked by relevance, importance, recency, and vector similarity.

• query — text to search against memory summaries
• tags — filter by tags
• related_user — filter by user/agent ID
• related_wallet — filter by Solana wallet
• memory_types — filter by type: episodic, semantic, procedural, self_model, introspective
• limit — max results (1-50, default 5)
• min_importance — minimum importance threshold (0-1)
• min_decay — filter out faded memories (0-1)
• track_access — update access timestamps (boolean)
• skip_expansion — skip LLM query expansion to save latency (boolean)

store_memory

Store a new memory. Memories persist across conversations and decay over time if not accessed.

• type — episodic (events), semantic (knowledge), procedural (behaviors), self_model (identity), introspective (journal)
• content — full memory content (max 5000 chars)
• summary — short summary for recall matching (max 500 chars)
• tags — tags for filtering
• concepts — structured labels (auto-inferred if omitted)
• importance — importance score 0-1 (auto-scored if omitted)
• emotional_valence — -1 (negative) to 1 (positive)
• source — origin identifier (e.g. mcp:my-agent)
• source_id — external ID (tweet ID, message ID, etc.)
• related_user — associated user/agent ID
• related_wallet — associated Solana wallet
• metadata — arbitrary JSON metadata

batch_store_memories

Store up to 50 memories in a single call. Returns an array of results with memory IDs. Same fields per memory as store_memory.

list_memories

Browse memories without a search query. Paginated; sort by recency, importance, or last access. Filter by type or tags.

update_memory

Update fields of an existing memory by its ID. Only provided fields are changed.

delete_memory

Permanently delete a memory by its ID. Irreversible.

get_memory_stats

Get statistics: counts by type, average importance/decay, dream session history, top tags.

find_clinamen

Anomaly retrieval — find high-importance memories with low relevance to current context. Useful for lateral thinking and unexpected connections.

• context — current context/topic (required)
• limit — max results (1-10, default 3)
• memory_types — filter by type
• min_importance — minimum importance (0-1, default 0.6)
• max_relevance — maximum relevance threshold (0-1, default 0.35)

Modes

The MCP server auto-detects its mode from environment:

Mode	Config	Storage
Hosted	CORTEX_API_KEY	clude.io (zero setup)
Self-hosted	SUPABASE_URL + SUPABASE_SERVICE_KEY	Your Supabase
Local SQLite (default)	none	~/.clude/brain.db (local embeddings, fully offline)
Local JSON	--local flag or CLUDE_LOCAL=true	~/.clude/memories.json (portable single file)

Setup

npx @clude/sdk setup

Or install manually:

npm install -g @clude/sdk
clude mcp-install

Architecture

• 5-tier memory: episodic (7%/day decay), semantic (2%/day), procedural (3%/day), self_model (1%/day), introspective (2%/day)
• Hybrid retrieval: pgvector cosine similarity + keyword matching + tag scoring
• Dream cycles: consolidation, compaction, reflection, contradiction resolution, emergence
• Association graph: typed bonds (causes, supports, contradicts, elaborates, resolves, etc.)
• Entity knowledge graph: auto-extracted people, projects, concepts, tokens, wallets
• Granular decomposition: per-fragment embeddings for sub-memory retrieval

License

MIT