Point any MCP client at uvx --from f1verse f1verse-mcp. There is no install step and no dependency tree: the server is standard library only, so it starts and answers tools/list in about 140 ms.
{
"mcpServers": {
"f1verse": {"command": "uvx", "args": ["--from", "f1verse", "f1verse-mcp"]}
}
}Eight tools, not eighty — a catalogue a model can actually choose from. The server tells the client to resolve vague references with f1_season_status first, and to check f1_data_quality before calling a fresh result final.
For any other LLM pipeline, the library emits the same catalogue itself: f1verse.tools() for MCP-dialect schemas, f1verse.tools("openai") for function-calling schemas, and f1verse.call_tool(name, arguments) to execute one. There is one catalogue behind both, so they cannot drift.
Errors are written for the caller that has to fix them without reading this page: an unknown tool name lists the real ones, a missing argument names it, an unexpected argument names what is accepted.