Summary

A working demo key means you can judge the results before you pay.

Most search APIs make you register before you can see whether the matches are any good; here the demo key returns real result structure with the values masked, so the evaluation costs nothing. The two tools split cleanly — one hunts for individuals, the other for companies and then the people inside them — which is the difference between recruiting and prospecting.

What it is

The MCP interface to Pearch.AI, a natural-language search over people and over B2B companies and leads. It runs locally over stdio, or you can point a client at the hosted endpoint; the same API key works for both.

What you get

`search_people` takes a request like "software engineers in California with 5+ years Python" and returns candidates with optional insights and profile scoring
`search_company_leads` finds companies and the contacts inside them — "AI startups in SF, 50–200 employees" plus "CTOs and engineering managers"
`search_people` also follows up on an existing thread rather than starting over
The demo key `test_mcp_key` returns masked sample results with no sign-up, so you can see the shape of the output first
Remote access over Streamable HTTP at https://mcp.pearch.ai/mcp, with `GET /health` for liveness

Requirements

Python 3.10+ and FastMCP (`pip install fastmcp` or `uv add fastmcp`); clone the repository and run `fastmcp install claude-desktop pearch_mcp.py --env PEARCH_API_KEY=test_mcp_key`, or paste the equivalent block into `~/.cursor/mcp.json` or `.vscode/mcp.json`. `PEARCH_API_KEY` holds either the demo key or a real key from the Pearch dashboard; `PEARCH_API_URL` overrides the base URL, which defaults to https://api.pearch.ai.

Setup effort

One command — fastmcp install claude-desktop pearch_mcp.py --env PEARCH_API_KEY=test_mcp_key