Trello

Fetches and caches Trello board hierarchies (boards → lists → cards) via the Trello REST API, maintaining a local representation of board structure that can be queried without repeated API calls. Implements MCP resource protocol to expose boards as queryable entities with lazy-loading of nested lists and cards, reducing API rate-limit pressure for frequent state checks.

Creates new cards in specified Trello lists by accepting card name, description, and list ID, then calling Trello's POST /cards endpoint with proper payload formatting. Supports optional parameters like due dates, labels, and assignees, with validation to ensure list exists before card creation to prevent orphaned cards.

Updates card properties (name, description, due date, labels, position, list membership) via Trello's PUT /cards/{id} endpoint with field-level validation and conflict detection. Implements optimistic updates with rollback capability if the API rejects changes due to concurrent modifications or invalid state transitions.

Retrieves all lists within a Trello board and exposes them as queryable resources with optional filtering by list name, ID, or status (open/closed). Uses Trello's GET /boards/{id}/lists endpoint with caching to avoid repeated enumeration, enabling agents to discover target lists dynamically without hardcoding list IDs.

Searches for cards across a board or within specific lists using criteria like card name, description content, labels, assignees, or due date ranges. Implements client-side filtering on top of Trello's GET /boards/{id}/cards endpoint since Trello API lacks server-side search, with optional caching to reduce API calls for repeated queries.

Retrieves available labels on a board and applies or removes labels from cards via Trello's PUT /cards/{id}/idLabels endpoint. Supports label creation if labels don't exist, with color and name validation to ensure labels conform to Trello's constraints.

Exposes board and list metadata (name, description, creation date, member list, permission settings) as MCP resources with read-only access. Implements caching to avoid repeated metadata fetches, enabling agents to understand board context and member structure without querying Trello API repeatedly.