MCP (Model Context Protocol) server for MemoClaw — semantic memory for AI agents. Store, recall, and manage memories with vector search. 100 free API calls per wallet, no registration needed.
Installation
npm install -g memoclaw-mcp
Configuration
Set your private key via environment variable or config file:
By default, the server uses stdio transport. To run as an HTTP server (MCP spec 2025-03-26 Streamable HTTP):
# Via CLI flag
memoclaw-mcp --http
# Via environment variable
MEMOCLAW_TRANSPORT=http memoclaw-mcp
# Custom port (default: 3100)
MEMOCLAW_PORT=8080 memoclaw-mcp --http
The server exposes:
POST /mcp — MCP Streamable HTTP endpoint
GET /health — Health check returning { status: "ok", version: "..." }
Configure your MCP client to connect to http://localhost:3100/mcp.
Rate Limiting
The HTTP transport includes built-in rate limiting to prevent abuse:
| Environment Variable | Default | Description |
|---------------------|---------|-------------|
| MEMOCLAW_RATE_LIMIT | 60 | Max requests per IP per window |
| MEMOCLAW_GLOBAL_RATE_LIMIT | 200 | Max total requests per window |
| MEMOCLAW_SESSION_RATE_LIMIT | 10 | Max new sessions per IP per window |
| MEMOCLAW_RATE_LIMIT_WINDOW_MS | 60000 | Window duration in ms (1 minute) |
| MEMOCLAW_TRUST_PROXY | false | Trust X-Forwarded-For header for client IP (set to true or 1 when behind a reverse proxy) |
Set any limit to 0 to disable it. Rate-limited requests receive a 429 Too Many Requests response with a Retry-After header.
Tools
Core
| Tool | Description |
|------|-------------|
| memoclaw_store | Store a memory with semantic embeddings |
| memoclaw_recall | Semantic search — find memories by meaning |
| memoclaw_search | Keyword search — find memories by exact text |
| memoclaw_get | Get a single memory by ID |
| memoclaw_list | List memories chronologically |
| memoclaw_delete | Delete a memory by ID |
| memoclaw_update | Update a memory's fields |
| memoclaw_context | Get contextually relevant memories using GPT-4o-mini |
Bulk operations
| Tool | Description |
|------|-------------|
| memoclaw_bulk_store | Store up to 100 memories in one call |
| memoclaw_bulk_delete | Delete up to 100 memories in one call |
| memoclaw_batch_update | Update up to 50 memories in one call |
| memoclaw_import | Import memories from a JSON array |
| memoclaw_export | Export all memories as JSON or JSONL |
Ingestion and extraction
| Tool | Description |
|------|-------------|
| memoclaw_ingest | Dump conversations/text, auto-extract facts with dedup and relations |
| memoclaw_extract | Extract structured facts from conversation via LLM |
| memoclaw_consolidate | Merge similar/duplicate memories by clustering |
| memoclaw_migrate | Bulk-import .md/.txt files into MemoClaw |
Organization
| Tool | Description |
|------|-------------|
| memoclaw_pin | Pin a memory to prevent decay |
| memoclaw_unpin | Unpin a memory |
| memoclaw_tags | List all unique tags with counts |
| memoclaw_namespaces | List all namespaces with counts |
| memoclaw_count | Count memories (with optional filters) |
| memoclaw_delete_namespace | Delete all memories in a namespace |
Knowledge graph
| Tool | Description |
|------|-------------|
| memoclaw_create_relation | Create a relationship between two memories |
| memoclaw_list_relations | List relationships for a memory |
| memoclaw_delete_relation | Delete a relationship |
| memoclaw_graph | Traverse the memory graph from a starting memory |
Insights and status
| Tool | Description |
|------|-------------|
| memoclaw_suggested | Get proactive suggestions (stale, fresh, hot, decaying) |
| memoclaw_core_memories | Get your most important/frequently accessed memories |
| memoclaw_stats | Memory usage statistics |
| memoclaw_history | View edit history for a memory |
| memoclaw_check_duplicates | Check if similar content exists before storing |
| memoclaw_status | Check free tier remaining calls |
| memoclaw_init | Verify configuration and API connectivity |
Example usage
Once configured, your AI agent can:
"Remember that the meeting is at 3pm tomorrow"
"What did I say about the project deadline?"
"Show me my core memories"
"Consolidate duplicate memories (dry run first)"
Pricing
Free Tier: Every wallet gets 100 free API calls — no payment required.
After the free tier, x402 micropayments kick in automatically: