# Trello MCP server

This is a read-only first version of a FastMCP server backed by the Trello REST API.

## Tools

- `list_boards`
- `find_board`
- `list_board_lists`
- `list_board_cards`
- `list_list_cards`
- `get_card`
- `get_card_members`
- `get_card_checklists`
- `get_completed_check_items`
- `get_board_activity`

The deployed server receives `TRELLO_API_KEY` and `TRELLO_TOKEN` from Google
Cloud Secret Manager through its Cloud Run configuration. These credentials
must not be stored in this repository or added to `.env.example`. MCP clients
only need the server URL and their Cloud Run authentication credentials.

## Test the remote MCP server

The private Cloud Run service can be tested with the included client. It uses
your local `gcloud auth print-identity-token` credentials and does not access
the Trello secrets.

```bash
cp .env.example .env
# Set MCP_URL in .env to your deployed /mcp endpoint.
python client.py
python client.py list_boards
python client.py find_board \
  --args-json '{"board_name":"project_architecture"}'
```

The ignored `.env` file supplies `MCP_URL`. You can still use `--url` to test
another deployment. In an environment without `gcloud`, provide a short-lived
token through `MCP_BEARER_TOKEN`.

## Run locally

Running the server itself locally still requires Trello credentials. Supply
them through your shell or a private, ignored `.env` file; never commit them.

```bash
pip install -e ".[dev]"
TRELLO_API_KEY=... TRELLO_TOKEN=... python server.py
```

The HTTP MCP endpoint is available at `http://localhost:8080/mcp`.

For a local stdio client, use FastMCP's CLI against `server.py` or change the
`mcp.run` transport in the `__main__` block to `stdio`.

## Deploy on Cloud Run

Build and deploy from this directory. Cloud Run supplies `PORT`; the container
binds to `0.0.0.0` and exposes the MCP endpoint at `/mcp`.

The service should be protected with Cloud Run authentication or an MCP
authentication layer before it is exposed publicly.