There is nothing to configure beyond the RapidAPI key, and the whole surface is a profile URL in, JSON out — which makes the third-party dependency the thing to weigh. Profile data comes from the Fresh LinkedIn Profile Data API on RapidAPI, so coverage, rate limits and cost are that provider's, not this server's.
A single-purpose MCP server that reads LinkedIn profile data. It takes a profile URL, calls the Fresh LinkedIn Profile Data API through RapidAPI, and returns the profile as JSON for the model to work with.
- Profile data for one LinkedIn URL, returned as JSON — `get_profile`
- Skills included, with most of the optional extra sections turned off, so the response stays to the profile itself
- Non-blocking calls through `httpx`, so a slow lookup does not stall the server
- A clear failure path: a missing key raises at startup, and a failed API call returns a message saying the profile could not be fetched rather than an empty result
A RapidAPI key in `RAPIDAPI_KEY`, read from a `.env` file in the project directory or from the environment. Python 3.7 or higher with the MCP framework, `httpx` and `python-dotenv` — the documented install is `uv add mcp[cli] httpx requests`. The server runs as `uv run linkedin.py` over stdio, and the client entry points at the project directory.