Cognee

Community

Memory manager with graph/vector stores, 30+ data sources.

About Cognee MCP Server

Cognee is an MCP server in the RAG & Knowledge category. It exposes typescript, local to connected AI applications, enabling them to interact with cognee capabilities through the standardized Model Context Protocol.

Features

TypeScript
Local

Category

RAG & Knowledge
Getting Started

To use the Cognee MCP server in your AI application, follow these steps:

  1. Install the MCP client SDK for your platform
  2. Connect to the Cognee MCP server
  3. Authenticate with your Cognee credentials
  4. Start using the tools, resources, and prompts provided by the server

Example Code


// Connect to the Cognee MCP server
const server = await mcp.connect("cognee");

// List available tools
const tools = await server.tools.list();

// List available resources
const resources = await server.resources.list();

// List available prompts
const prompts = await server.prompts.list();
                      
Installation

npm

npm install @mcp/server-cognee

pip

pip install mcp-server-cognee
Compatible With
Claude Desktop
Cursor
Windsurf
Zed
Any MCP-compatible client