Documentation

Quick Start

Get memory working in under a minute. No signup required for local mode.

npx @lakehouse/memory-mcp

This starts the MCP server with local storage. Your memories are saved to ~/.lakehouse42/memory-mcp/memories.json

Claude Code

Add memory to Claude Code by editing your config file.

Step 1: Edit config

Open ~/.claude/claude_code_config.json

{
  "mcpServers": {
    "memory": {
      "command": "npx",
      "args": ["@lakehouse/memory-mcp"]
    }
  }
}

Step 2: Restart Claude Code

Close and reopen your terminal. The memory tools will now be available.

Claude Desktop

Add memory to Claude Desktop via the settings menu.

1. 1Open Claude Desktop Settings
2. 2Go to Developer → Edit Config
3. 3Add the MCP server configuration

{
  "mcpServers": {
    "memory": {
      "command": "npx",
      "args": ["@lakehouse/memory-mcp"],
      "env": {
        "LH42_URL": "https://api.lakehouse42.com",
        "LH42_API_KEY": "lh42_your_api_key"
      }
    }
  }
}

Cursor / Windsurf

Add memory to Cursor or Windsurf following the same pattern.

{
  "mcpServers": {
    "memory": {
      "command": "npx",
      "args": ["@lakehouse/memory-mcp"]
    }
  }
}

LH42 Backend

Connect to LH42 for full semantic search, deduplication, and knowledge graph features.

Get your API key

Option 1: Web Signup

1. Sign up for LH42
2. Go to Settings → API Keys
3. Create a new API key

Option 2: Programmatic Registration (for AI agents)

# Any provider - embeddings handled automatically
curl -X POST https://api.lakehouse42.com/api/v1/agents/register \
  -H "Content-Type: application/json" \
  -d '{
    "agent_id": "my-agent",
    "llm_provider": "anthropic",
    "llm_api_key": "sk-ant-xxx"
  }'

# Returns: { "api_key": "lh_...", "embedding_provider": "platform (managed bge-m3)" }

No existing API key required. Supported LLM providers: openai, anthropic, google, cohere, voyage, azure-openai. Anthropic and other providers without native embeddings automatically use platform-managed BGE-M3 embeddings.

Add environment variables

{
  "mcpServers": {
    "memory": {
      "command": "npx",
      "args": ["@lakehouse/memory-mcp"],
      "env": {
        "LH42_URL": "https://api.lakehouse42.com",
        "LH42_API_KEY": "lh42_your_api_key"
      }
    }
  }
}

MCP Tools Reference

Unified Search

Search across both your personal memories AND enterprise documents with a single query. Results are merged using Reciprocal Rank Fusion (RRF) for optimal relevance.

Memory Types

REST API

Access memory and knowledge search directly via the LH42 REST API.

Memory Operations

# Store a memory
curl -X POST https://api.lakehouse42.com/api/v1/agents/memory \
  -H "Authorization: Bearer lh_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{"content": "User prefers TypeScript", "type": "preference"}'

# Recall memories
curl -X POST https://api.lakehouse42.com/api/v1/agents/memory/recall \
  -H "Authorization: Bearer lh_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{"query": "programming preferences", "limit": 10}'

Unified Knowledge Search

Search both personal memories AND enterprise documents in a single call.

# Unified search (memories + documents)
curl -X POST https://api.lakehouse42.com/api/v1/agents/knowledge/search \
  -H "Authorization: Bearer lh_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{
    "query": "project deadlines",
    "include_memories": true,
    "include_documents": true,
    "memory_weight": 0.4,
    "limit": 10
  }'

# List available document collections
curl https://api.lakehouse42.com/api/v1/agents/knowledge/collections \
  -H "Authorization: Bearer lh_your_api_key"

Programmatic Usage

Use the package in your own TypeScript/JavaScript code.

import { createMemoryServer, LH42Backend } from "@lakehouse/memory-mcp";

// Create server with custom config
const server = await createMemoryServer({
  lh42Url: "https://api.lakehouse42.com",
  apiKey: "lh42_xxx",
  debug: true,
});

// Or use backends directly
const backend = new LH42Backend({
  url: "https://api.lakehouse42.com",
  apiKey: "lh42_xxx",
});

await backend.initialize();
await backend.remember({ content: "User likes TypeScript" });
const results = await backend.recall({ query: "programming preferences" });