What is the Basketball API MCP server?

It is a Model Context Protocol server that gives AI assistants 15 tools to read basketball data: games, box scores, play by play, shots, standings, team form and player splits. You connect it once with your API key, then ask questions in plain words and the assistant answers from the data.

Which assistants work with it?

Any assistant or tool that supports remote MCP servers over Streamable HTTP and lets you set a header, including Claude Code, Cursor and VS Code. The connect section above has the exact setup for each.

Which plan do I need?

The MCP server is on the Complete plan. Your key on another plan still works with the REST API; the server tells the assistant which plan it needs, and the assistant can pass that on.

No. Each tool call is one API call on your key and counts towards your plan's daily calls, like any other call. The usage tool does not count.

Is the data the same as the REST API?

Yes. Every tool forwards to the REST API with your key, so the answers, ids, fields and limits are exactly the REST API's. Anything you find through an assistant you can fetch again in your own code.

As current as the API: results land soon after the final whistle, box scores, play by play and shots follow.

Can an assistant change anything on my account?

No. Every tool only reads. Keys, webhooks and your account can only be changed on your dashboard or through the REST API.

What happens when I reach my limit?

The tool answers with a short message saying you are going too fast or today's calls are used up, and when they reset. The connection itself keeps working.

Basketball MCP server

Ask in plain words. Your assistant calls the API.

The Basketball API MCP server connects an AI assistant to live and historical basketball data: scores, box scores, play by play, shot charts, standings, team form and player splits. Ask a question the way you would ask a colleague, and the assistant picks the right call, reads the answer and replies with real numbers instead of guesses.

https://basketballapi.net/mcp

On the Complete plan. Every tool call is one API call on your key.

Toolget_team_form

What an MCP server is, and why it matters for basketball data

MCP, the Model Context Protocol, is an open standard that lets an AI assistant use outside tools while it answers you. An MCP server lists the tools it offers, each with a name, a plain description and the inputs it takes. The assistant reads that list, decides which tool fits your question, calls it and works from what comes back.

Without it, an assistant answers basketball questions from whatever it remembers, which is often out of date and sometimes simply wrong: last night's scores, a player's shooting this month or a team's home record are not things a model can know. With the Basketball API MCP server, those answers come from the same database our REST API serves, checked and updated every few minutes, so the numbers in the reply are the numbers in the game.

You do not write code to use it. Add one address and your API key to an assistant that supports MCP, and ask.

Connect your assistant in a minute

The server speaks Streamable HTTP and takes your key in the Authorization header, the same key you use for the REST API. Your key is on your dashboard. Pick your assistant for the exact setup. For agents that read skills, the agent skill explains the whole API in one file.

What you can ask

Anything the API can answer, in your own words. These are the kinds of questions people ask, and the tool the assistant reaches for.

The 15 tools

Each tool is one endpoint of the REST API, with the same plans, limits and answers, so anything you learn through an assistant you can build on in code. The assistant sees a short description of each; here is what each one is for.

How a question becomes an answer

What happens between your question and the reply, step by step.

The assistant reads the tool list.When you connect, your assistant asks the server which tools it offers. It gets the 15 tools with their names, what they do and the inputs each takes.
It picks a tool and the inputs.From your question it chooses a tool, say get_games for "who plays tonight in the EuroLeague", and fills in the date and competition.
The call runs on your key.The server passes the call to the REST API with your key, so your plan, your limits and your usage page all count it, exactly as if your own code had made it.
It chains calls when it needs to.A question about a player's shooting last night may take two calls: get_games to find the game id, then get_shots for that game. The assistant follows the ids from one answer to the next.
It answers from the data.The reply is written from the JSON that came back: real scores, real lines and real percentages, which you can check against the same call in the REST API.

Who uses it, and for what

League tables and form

Ask where a team sits in its league, how it does at home and away, or how it has played over its last ten games. get_standings and get_team_form give the table and the run, with pace and ratings.

Fantasy and player analysis

Check a player's last ten games, how his scoring changes at home and away or off the bench, and his true shooting and usage. get_player_games and get_player_splits give game-by-game lines and averages without a spreadsheet.

Writing and broadcasting

Get the facts behind a story while you write it: how a lead changed, who took the late shots, a team's pace and ratings over a run of games. Every figure traces back to one call you can run again.

Scouting and coaching

Look at where a team's shots come from, how often they score at the rim, and how a rotation's minutes are spread. Shot positions are in feet from the basket, so charts and zones are yours to draw.

Building apps faster

Developers use the MCP server to explore the data before writing code: ask, look at the JSON, then make the same call from your app. The ids, fields and errors are the REST API's own.

Everyday questions

Last night's scores in a league you follow, your team's next games, or how a rookie has started his season: one sentence in your assistant instead of three websites.

The data behind every answer

The MCP server reads the same data as the REST API, across professional and international basketball.

Good to know

Your key, your limitsEvery tool call goes through the REST API on your key, so it counts towards your daily calls like any other and shows on your usage page.

Refusals in wordsA call your plan does not reach, or one made too fast, comes back as a short message the assistant can read and pass on, never as a broken connection.

Same ids everywhereIds from a tool work in the REST API and the other way round: cp_ competitions, ss_ seasons, tm_ teams, gm_ games, pl_ players.

Read onlyThe tools only read. Nothing an assistant does through them can change your account, your keys or your webhooks.

Times in UTCDays are the venue's own calendar day as YYYY-MM-DD, and every time is UTC, so an assistant never mixes up time zones.

Your key stays yoursKeep the key in your assistant's settings, never in a shared chat. If one leaks, revoke it on your dashboard and calls with it stop at once.

Questions about the MCP server

The MCP server is on Complete

Every endpoint, every season and the highest limits, with the MCP server included.

Read the documentation See the plans