Don’t see your tool? Cognee MCP works with any client that speaks the Model Context Protocol — not only the ones listed below. Coding tools such as OpenCode, Windsurf, and Zed can connect too. If your client supports remote or HTTP MCP servers, point it at your running Cognee MCP server’s HTTP URL (
http://localhost:8000/mcp). If it expects a local stdio server, such as Claude Desktop, configure it to launch Cognee MCP over stdio — either the PyPI release via uvx cognee-mcp@latest or a source checkout. Follow the integration guide closest to your client’s config format.Available Clients
Cursor
AI-powered code editor with native MCP support
Claude Code
Command-line AI assistant from Anthropic
Claude Desktop
Native desktop app from Anthropic for macOS and Windows
Codex
OpenAI coding agent with MCP support
Cline
VS Code extension for AI-assisted development
Continue
Open-source AI coding assistant for VS Code and JetBrains
Python Agent
Connect your own Python LLM agent via stdio or HTTP
Configuration Options
Each integration requires a configuration file that tells the client how to connect to your Cognee MCP server. You have two options:Docker (HTTP Transport)
Use this if you started Cognee MCP with Docker using the quickstart guide. The client connects to the server over HTTP.Local (stdio Transport)
Use this if you run the server on the same machine as the client. The client starts the server as a subprocess and communicates through standard input/output. To run the released package without cloning anything, replace thecommand and args below with "command": "uvx" and "args": ["cognee-mcp@latest"]. The form shown here runs a source checkout: