integrations-mcp

Built by an autnomous agent over about a day

A comprehensive MCP (Model Context Protocol) server that exposes 390+ tools across 66 providers, giving AI assistants direct access to popular SaaS platforms, public APIs, and utility services.

Quick Start

# Install dependencies
pip install -e .

# Run the MCP server
python server.py

Configure provider-specific environment variables (see table below) for authenticated services. Providers with missing credentials are skipped gracefully at startup.

Architecture

server.py                  → FastMCP entry point
providers/__init__.py      → Lazy provider registry (all providers loaded on demand)
providers/<name>.py        → Individual provider modules, each exporting a register(mcp) function

Each provider is imported lazily so a missing optional dependency only disables that one provider — the rest of the server continues to work.

Provider Table

Total: 66 providers · 390+ tools · 34 authenticated · 32 free/public

Detailed Tool Reference

See providers/PROVIDERS.md for a full reference of every tool with parameters, descriptions, and example outputs.

Adding a New Provider

1. Create providers/your_provider.py with a register(mcp) function.
2. Use @mcp.tool() to decorate each tool function.
3. Add ("providers.your_provider", "register") to the _PROVIDERS list in providers/__init__.py.
4. Add any new dependencies to pyproject.toml.

def register(mcp):
    @mcp.tool()
    def your_tool_name(param: str) -> dict:
        """Description of what this tool does."""
        # implementation
        return {"result": "..."}

License

See repository for license details.