Feyod MCP Server

Feyod is a Model Context Protocol (MCP) server that turns natural‑language questions about Feyenoord football into precise SQL queries against a curated SQLite database. By exposing a single endpoint, the server allows AI assistants—such as Claude—to ask about match results, player statistics, line‑ups, cards, and opponent performance without requiring the user to write any code. The service is especially valuable for developers building sports analytics tools, chatbots, or data‑driven applications that need instant access to historical Feyenoord match information.

The core value of the server lies in its natural‑language interface. Internally, it uses LangChain to parse a user’s question into SQL. This pipeline includes optional few‑shot prompting for higher accuracy, automatic validation of the generated query, and a fallback that lets an LLM correct syntax errors before execution. Once validated, the SQL runs against the embedded Feyenoord database and the raw results are returned to the client. Because all heavy lifting is performed on the server, developers can focus on integrating the results into their workflows rather than managing query logic or database connections.

Key capabilities include:

• Dynamic LLM loading – the server can switch between providers such as OpenAI or Google’s Gemini by simply changing environment variables, giving teams flexibility in cost and performance.
• Tool discovery – via the MCP endpoint, clients can see that the server offers a single tool, , which accepts any natural‑language query about matches, players, or opponents.
• Public and containerized deployment – a ready‑to‑use public endpoint is available at , and a Docker image on Hub allows self‑hosting with local database mounts for custom datasets or offline use.

Typical use cases span sports journalism, fan engagement bots, and data‑science prototypes. A newsroom could embed the server in a conversational UI to let reporters quickly pull up match statistics, while a mobile app could answer user queries about their favorite player’s performance in real time. In research, data scientists can prototype SQL analyses by simply asking questions instead of writing queries, accelerating exploratory data analysis.

Because Feyod is built on an open‑source database maintained in a separate GitHub repo, developers can extend or modify the schema without altering the MCP logic. The combination of a lightweight SQLite backend, automated query generation, and seamless LLM integration makes Feyod a standout MCP server for sports analytics and any domain where structured data can be queried through natural language.