Lets an AI assistant find, run and build workflows on your n8n instance.

Cognee MCP
MCP server by Cognee (topoteretes)
Cognee MCP at a glance
Cognee MCP is an official MCP server from Cognee (topoteretes). It lets an AI assistant remember data in a knowledge graph and recall it with graph-aware search.
- Connects to
- Memory and automation
- Hosting
- Local · Streamable HTTP and stdio
- Sign-in
- API key
- Price
- Free
- Works with
- Claude Code, Cursor, Cline
- Best for
- Developers building agents with graph-based memory
- Checked
- October 3, 2026 on official sources
How to install Cognee MCP
Add the command below as a new MCP server in your AI app’s settings or MCP config file. Sign-in: API key.
docker run -e TRANSPORT_MODE=sse --env-file ./.env -p 8000:8000 cognee/cognee-mcp:mainWhat Cognee MCP can do
It runs Cognee's memory engine as an MCP server that turns text and files into a searchable knowledge graph. Agents can store data, recall it later and forget datasets they no longer need.
Best for: developers building agents with graph-based memory
Pros
- Open source and self-hostable
- Supports HTTP, SSE and stdio
- Can also point at Cognee Cloud
Cons
- Needs an LLM API key such as OpenAI
- Setup takes more steps than a single npx command
Cognee MCP FAQ
Is Cognee MCP free?
Yes. The server is open source under Apache-2.0, though you pay your own LLM provider for processing.
How do I connect Cognee MCP?
Run the Docker image or the Python server and point your client at it, for example claude mcp add cognee-sse -t sse http://localhost:8000/sse.
What can Cognee MCP do?
It stores text and files as graph memory, recalls them with search and deletes data by dataset or ID.