Trello

Translates natural language queries into structured Trello API calls by parsing user intent through an MCP tool registry that maps semantic requests to specific Trello REST endpoints. The server maintains a layered architecture with a Trello API client that handles authentication via API key/token, request formatting, and response normalization, allowing AI assistants to execute Trello operations without direct API knowledge.

Supports two distinct operational modes controlled via environment configuration: Claude App Mode (direct FastMCP integration with Claude Desktop via stdio) and SSE Server Mode (standalone HTTP server with Server-Sent Events for Cursor and other MCP clients). This dual-mode architecture allows the same codebase to serve both tightly-integrated desktop clients and distributed web-based clients without code branching.

Provides read-only traversal of Trello's hierarchical entity model (Boards → Lists → Cards → Checklists) through dedicated MCP tools that query the Trello API and return structured data about the full hierarchy. Each level supports filtering and detailed inspection, allowing AI assistants to understand board structure before performing mutations.

Enables creation and modification of Trello cards through MCP tools that accept natural language parameters (title, description, due date, labels) and translate them into Trello API PATCH/POST requests. Supports updating card attributes like name, description, due dates, and list assignment, with automatic validation of input parameters before API submission.

Provides operations to create, rename, and archive lists within a Trello board through MCP tools that map to Trello's list endpoints. Supports creating new lists with initial names, updating list names, and archiving (soft-deleting) lists without affecting cards. Implements list position management for reordering columns.

Enables creation, updating, and deletion of checklists and checklist items within cards through MCP tools that interact with Trello's checklist endpoints. Supports adding checklists to cards, creating checklist items, marking items as complete/incomplete, and managing item state without modifying the card itself.

Implements a tool registry that defines MCP tool schemas for all Trello operations (board queries, card creation, list management, etc.) with JSON schema validation for parameters. The registry maps natural language tool invocations to specific Python functions and validates inputs before execution, providing AI assistants with discoverable, self-documenting APIs for Trello operations.

Provides a Python wrapper around the Trello REST API that handles authentication (API key/token), request formatting, error handling, and response normalization. The client abstracts away HTTP details and Trello-specific conventions (e.g., URL construction, parameter encoding) and provides typed methods for common operations, reducing boilerplate in the service layer.

Implements a service layer that orchestrates Trello API client calls to provide higher-level operations (e.g., 'get all cards in a list', 'create a card with labels'). The service layer handles multi-step operations, data transformation, and error propagation, allowing the tool registry to invoke semantic operations without managing API call sequences.

Manages server configuration through environment variables (.env file) that control operational mode (Claude App vs SSE), Trello API credentials, and deployment settings. Supports both local development (.env.example) and containerized deployment (Docker Compose) configurations, with environment variable validation at server startup.