local-only server

The server can only run on the client’s local machine because it depends on local resources.

cognee MCP server

Installing Manually

A MCP server project

Clone the cognee repo
Install dependencies

brew install uv

cd cognee-mcp
uv sync --dev --all-extras --reinstall

Activate the venv with

source .venv/bin/activate

Add the new server to your Claude config:

The file should be located here: ~/Library/Application\ Support/Claude/

cd ~/Library/Application\ Support/Claude/

You need to create claude_desktop_config.json in this folder if it doesn't exist Make sure to add your paths and LLM API key to the file bellow Use your editor of choice, for example Nano:

nano claude_desktop_config.json

{
	"mcpServers": {
		"cognee": {
			"command": "/Users/{user}/cognee/.venv/bin/uv",
			"args": [
        "--directory",
        "/Users/{user}/cognee/cognee-mcp",
        "run",
        "cognee"
      ],
      "env": {
        "ENV": "local",
        "TOKENIZERS_PARALLELISM": "false",
        "LLM_API_KEY": "sk-"
      }
		}
	}
}

Restart your Claude desktop.

Installing via Smithery

To install Cognee for Claude Desktop automatically via Smithery:

npx -y @smithery/cli install cognee --client claude

Define cognify tool in server.py Restart your Claude desktop.

To use debugger, run:

mcp dev src/server.py

Open inspector with timeout passed:

http://localhost:5173?timeout=120000

To apply new changes while developing cognee you need to do:

poetry lock in cognee folder
uv sync --dev --all-extras --reinstall
mcp dev src/server.py

Development

In order to use local cognee build, run in root of the cognee repo:

poetry build -o ./cognee-mcp/sources

After the build process is done, change the cognee library dependency inside the cognee-mcp/pyproject.toml from

cognee[postgres,codegraph,gemini,huggingface]==0.1.38

to

cognee[postgres,codegraph,gemini,huggingface]

After that add the following snippet to the same file (cognee-mcp/pyproject.toml).

[tool.uv.sources]
cognee = { path = "sources/cognee-0.1.38-py3-none-any.whl" }

You must be authenticated.

A

security – no known vulnerabilities

A

license - permissive license

A

quality - confirmed to work

Tools

Memory manager for AI apps and Agents using various graph and vector stores and allowing ingestion from 30+ data sources