MCP Trello

# CLAUDE.md This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository. ## Common Commands ### Development - `npm run dev` - Start development server with HTTP streaming interface and hot reloading - `npm run dev:stdio` - Start development with TypeScript watch mode and MCP inspector (legacy) - `npm run watch` - TypeScript watch mode only - `npm run inspector` - Run MCP inspector for debugging tools ### Build Commands - `npm run build` - Build HTTP streaming version (default) - `npm run build:http` - Build HTTP streaming version using Smithery - `npm run build:stdio` - Build stdio version for backwards compatibility ### Deployment - `npm start` - Start HTTP streaming server (default) - `npm run start:http` - Start HTTP streaming server - `npm run start:stdio` - Start stdio server (backwards compatibility) ### Quality Assurance - `npm run lint` - Lint code with Biome - `npm run lint:fix` - Lint and auto-fix issues - `npm run format` - Check code formatting with Biome - `npm run format:fix` - Format code with Biome ### Testing/Debugging - `npm run debug` - Start server with Node.js inspector - `npm run debug:watch` - Debug mode with TypeScript watch - `npm run logs` - Tail Claude MCP logs ## Architecture This is a Model Context Protocol (MCP) server that provides Trello integration tools for AI assistants. It supports both HTTP streaming and stdio interfaces. The architecture follows a clean separation of concerns: ### Core Components **Main Server (`src/index.ts`)** - Exports factory function for HTTP streaming interface - Maintains backwards compatibility with stdio transport - Environment variable validation for Trello credentials - Configuration schema using Zod for type safety - Graceful shutdown handling for both interfaces **Trello Client (`src/trello/client.ts`)** - Axios-based HTTP client with built-in rate limiting - Tool registration method that defines all available Trello operations - Error handling with retry logic for rate limits - Type-safe API interactions using Zod schemas **Rate Limiting (`src/trello/rate-limiter.ts`)** - Implements Trello's rate limits (300 requests/10s per API key, 100/10s per token) - Token bucket algorithm for request throttling **Type Definitions (`src/trello/types.ts`)** - TypeScript interfaces for Trello API entities (cards, lists, actions, etc.) - Configuration interfaces for client setup ### Available Tools The server exposes these tools for Trello interaction: - Card management: `getCardsByList`, `getMyCards`, `addCard`, `updateCard`, `moveCard`, `archiveCard`, `changeCardMembers` - List management: `getLists`, `addList`, `archiveList` - Board information: `getRecentActivity` ## Environment Variables Required for operation: - `trelloApiKey` - Trello API key from https://trello.com/app-key - `trelloToken` - Trello token generated using API key - `trelloBoardId` - Target board ID from Trello board URL ## Build System - **TypeScript**: ES2020 target with Node16 module resolution - **Biome**: Linting and formatting with strict rules (no explicit any, unused variables as errors) - **Smithery**: HTTP streaming interface bundler - **Outputs**: - `build/` directory with TypeScript compilation and type declarations (stdio) - `.smithery/index.cjs` bundled version for HTTP streaming - **Entry Points**: - `build/index.js` (stdio executable) - `.smithery/index.cjs` (HTTP streaming)