Summary
Create a Model Context Protocol (MCP) server that exposes Twenty CRM's GraphQL and REST APIs as MCP tools, enabling AI assistants like Claude to directly interact with CRM data.
Goals
- Enable direct CRM management through Claude Code interface
- Production-ready MVP following Anthropic MCP standards
- Docker deployment with DXT packaging
- Full CRM access (CRUD, workflows, AI features)
- Token-efficient API design
Technical Approach
- Use official MCP TypeScript SDK (@modelcontextprotocol/sdk)
- Hybrid GraphQL + REST API integration
- Containerized deployment
- Multi-tenant workspace support
Implementation Tasks
Based on and :
- Project Setup - MCP server foundation with TypeScript
- Authentication - API key integration with Twenty's auth system
- API Integration - GraphQL and REST clients with connection pooling
- MCP Tools - CRUD, search, workflow, and AI agent tools
- Deployment - Docker container with DXT packaging
Acceptance Criteria
Related Documents
- PRD:
- Tasks:
- Development: directory created
This implementation leverages Twenty's existing robust API infrastructure to provide seamless AI integration without UI complexity.
Summary
Create a Model Context Protocol (MCP) server that exposes Twenty CRM's GraphQL and REST APIs as MCP tools, enabling AI assistants like Claude to directly interact with CRM data.
Goals
Technical Approach
Implementation Tasks
Based on and :
Acceptance Criteria
Related Documents
This implementation leverages Twenty's existing robust API infrastructure to provide seamless AI integration without UI complexity.