Persistent Memory via MCP

Give Claude Desktop, Claude Code, Cursor, and any MCP-compatible AI tool long-term memory — with a single config change and zero code.

What is MCP? Model Context Protocol is the emerging standard for connecting AI tools to external services. An MCP server exposes "tools" that AI assistants can call natively — no SDK, no wrapper, no glue code.

Available Tools

The 0Latency MCP server exposes six tools that cover the full memory lifecycle:

memory_add

Extract and store memories from conversation turns. Entities, facts, and preferences are auto-extracted.

memory_recall

Recall relevant memories for the current context. Returns a token-budgeted context block.

memory_search

Full-text search across stored memories. Filter and rank by relevance.

memory_list

List memories with type filters and pagination. Browse what your agent remembers.

memory_delete

Remove specific memories by ID. Full control over what persists.

memory_graph

Query the knowledge graph — entities, relationships, and connection paths.

Quick Start

1. Get Your API Key

2. Install

# Clone and build
git clone https://github.com/0latency/mcp-server.git
cd mcp-server
npm install && npm run build

# Or install globally (when published to npm)
npm install -g @0latency/mcp-server

3. Configure Claude Desktop / Claude Code

Add to ~/.claude/claude_desktop_config.json or your project's .mcp.json:

{
  "mcpServers": {
    "0latency": {
      "command": "node",
      "args": ["/path/to/mcp-server/dist/index.js"],
      "env": {
        "ZERO_LATENCY_API_KEY": "zl_live_your_key_here"
      }
    }
  }
}

If installed globally via npm, simplify to:

{
  "mcpServers": {
    "0latency": {
      "command": "0latency-mcp",
      "env": {
        "ZERO_LATENCY_API_KEY": "zl_live_your_key_here"
      }
    }
  }
}

4. Configure Cursor

Add to .cursor/mcp.json in your project root:

{
  "mcpServers": {
    "0latency": {
      "command": "node",
      "args": ["/path/to/mcp-server/dist/index.js"],
      "env": {
        "ZERO_LATENCY_API_KEY": "zl_live_your_key_here"
      }
    }
  }
}

Environment Variables

Variable	Required	Default	Description
`ZERO_LATENCY_API_KEY`	✅	—	Your 0Latency API key
`ZERO_LATENCY_API_URL`	—	`https://api.0latency.ai`	API base URL (for self-hosted deployments)

How It Works

Once configured, your AI assistant can use 0Latency tools natively. No SDK needed — the MCP protocol handles everything.

Example: Say "Remember that I prefer TypeScript and dark mode" — the assistant calls memory_add. Next session, ask "What are my preferences?" — the assistant calls memory_recall and gets the answer instantly.

The knowledge graph lets your AI explore relationships between entities, find connection paths, and build a rich understanding of your project over time.

Source Code

The MCP server is open source and available on GitHub. Contributions welcome.

Other integrations: OpenClaw · LangChain · CrewAI · AutoGen