Overview
What is Pipe-MCP?
Pipe-MCP is a simple Model Context Protocol (MCP) server implementation for building AI assistants with specialized tools. It integrates with any MCP-compatible client (e.g., Claude Desktop, Claude.ai) and provides built‑in Pipedrive CRM tools.
How to use Pipe-MCP?
Clone the repository, install dependencies with uv, and configure environment variables (e.g., PIPEDRIVE_API_TOKEN). Run the server using uv run src/main.py (local) or via Docker. Connect an MCP client by providing a JSON configuration that specifies either SSE or stdio transport. The server supports both SSE (web) and stdio (CLI) transports.
Key features of Pipe-MCP
- Simple MCP server implementation using FastMCP
- Context management and lifecycle handling
- Demo echo tool for testing
- Integrated Pipedrive CRM API tools
- Support for both SSE and stdio transports
Use cases of Pipe-MCP
- Build AI assistants that can read and write Pipedrive deals
- Search and manage Pipedrive contacts and organizations via natural language
- Create custom tools for MCP-compatible clients (e.g., Claude Desktop)
- Prototype and test MCP server features with the built‑in echo tool
FAQ from Pipe-MCP
What transports does Pipe-MCP support?
Pipe-MCP supports both SSE (for web clients) and stdio (for CLI integration or MCP clients like Claude Desktop).
What runtime dependencies are required?
Python 3.12+, uv (recommended for package management), and optional Docker for containerized deployment.
How do I authenticate with Pipedrive?
Set the environment variable PIPEDRIVE_API_TOKEN with your Pipedrive API token. Obtain the token from your Pipedrive account.
Where does data live?
Pipedrive data lives in the Pipedrive cloud. The server configuration (host, port, transport) is set via environment variables. No local database is mentioned.
Can I add my own tools?
Yes. Define tool functions in src/tool.py (or a new module) and register them in src/server.py using the @mcp.tool() decorator.
