Trello MCP

Trello MCP connects an agent to Trello boards through the Model Context Protocol. Trello ships an official cloud-hosted server, and several community implementations built on the Trello REST API have existed longer and go wider, so the exact tool list depends on which one you point your client at. All of them map the same object model: boards contain lists, lists contain cards, and cards carry members, labels, due dates, checklists, comments and attachments. A typical server lets an agent enumerate the boards it can see, read the lists and cards on one, search across cards, create a card in a named list, move a card between lists, and update its fields. Richer community builds add checklists, comments, attachments, custom fields, member assignment and workspace-level operations, which is what you need if the goal is to run a board rather than only read it. The obvious use is turning unstructured input into tracked work: an agent reads a meeting note or a support thread and files the resulting cards in the right list with owners and due dates, or moves cards along a pipeline as it completes steps. Reading matters too, since a board-state summary is a cheap standup input. Access is an API key and token, or an OAuth grant for the hosted server, and the agent inherits exactly the board access of that account, so put it on a limited user if it should not see every workspace.