Installation¶

documentdb-mcp is a standard Python package and a prebuilt container image. Pick the path that matches how you want to run it.

Requirements¶

Python 3.11 – 3.14.
A reachable DocumentDB or MongoDB-compatible endpoint — see Backing Platform to deploy one locally.

From PyPI (recommended)¶

pip install documentdb-mcp

Optional extras¶

The base install ships the MCP server. Install the extra for what you need:

Extra	Install	Pulls in
(base)	`pip install documentdb-mcp`	MCP-server runtime (`agent-utilities[mcp]`, `pymongo`)
`agent`	`pip install "documentdb-mcp[agent]"`	Pydantic-AI agent + Logfire tracing (`agent-utilities[agent,logfire]`)
`all`	`pip install "documentdb-mcp[all]"`	MCP server, agent, and Logfire tracing
`test`	`pip install "documentdb-mcp[test]"`	`pytest`, `pytest-asyncio`, `pytest-cov`, `pytest-xdist`

# Typical: run the MCP server and the integrated agent
pip install "documentdb-mcp[all]"

From source¶

git clone https://github.com/Knuckles-Team/documentdb-mcp.git
cd documentdb-mcp
pip install -e ".[all]"          # editable install with every extra

With uv:

uv pip install -e ".[all]"
uv run documentdb-mcp

Prebuilt Docker image¶

A multi-stage, slim image is published on every release (entrypoint documentdb-mcp):

docker pull knucklessg1/documentdb-mcp:latest

docker run --rm -i \
  -e MONGODB_URI=mongodb://your-documentdb:27017/ \
  knucklessg1/documentdb-mcp:latest        # stdio transport (default)

For an HTTP server with a published port and the agent server, see Deployment.

Verify the install¶

documentdb-mcp --help
python -c "import documentdb_mcp; print(documentdb_mcp.__version__)"

Next steps¶

Deployment — run it as a long-lived MCP server and agent behind Caddy + DNS.
Usage — call the tools, the API, and the CLI.
Configuration — every environment variable.