Trello MCP Server

What is Trello MCP Server?

A Model Context Protocol (MCP) server that provides a natural language interface to the Trello API. It allows LLMs to read and modify Trello boards, lists, cards, and members, with an optional Notion integration for publishing cards.

How to use Trello MCP Server?

Install dependencies with uv pip install -r uv.lock, configure a .env file with your Trello API key and token, then create a mcp.json file pointing to the server script. Start the server through any MCP client that reads the mcp.json configuration.

Key features of Trello MCP Server

• Get all boards and board details
• Get lists inside a board
• Get cards inside a list and card details
• Create, update, and move cards between lists
• Add comments to cards
• Get board members
• Optional integration to publish cards to Notion

Use cases of Trello MCP Server

• Query Trello boards and card details via natural language
• Automate card creation and updates across multiple boards
• Move cards between lists (e.g., update status)
• Add comments to cards through a chat interface
• Publish Trello card content to a Notion page (with optional setup)

FAQ from Trello MCP Server

What dependencies are required?

Python and uv are required. You must have a Trello API key and token (obtained from trello.com/power-ups/admin). The Notion integration is optional.

How do I configure the server?

Create a .env file with TRELLO_API_KEY and TRELLO_API_TOKEN. Optionally add NOTION_API_KEY and NOTION_PARENT_PAGE_ID. Then configure a mcp.json with the path to uv and the server directory.

Can I use this server without Notion?

Yes. The Notion environment variables and integration are entirely optional. Leave them unset to use only Trello features.

What tools does the server expose?

Tools include: get all boards, get board details, get lists in a board, get cards in a list, get card details, create card, update card, move card, add comment, get board members, and publish to Notion.

How do I start the server?

The server is started via an MCP client that reads the mcp.json configuration. There are no separate start commands; the client invokes uv run server.py with the provided environment variables.