Nba MCP Server

What the labeveryday/nba_mcp_server MCP server does

The labeveryday/nba_mcp_server MCP server gives MCP clients access to live and historical NBA information. It covers player profiles, season and career statistics, game results, box scores, schedules, standings, awards, shooting data, team rosters, and advanced metrics.

Many tools accept a player or team name directly, so an agent does not need to look up an ID first. Resolution tools are also available when an application needs the matching player or team identifier and, for players and teams, related public NBA CDN image URLs.

How it works

The server retrieves information from the NBA’s live data, statistics, and schedule APIs. It can run over the default local stdio transport or as a streamable HTTP or SSE server. The documented standalone commands expose the same MCP server for local testing or client configuration.

Tool results are returned as one JSON object inside MCP text content. The object includes the invoked tool, its arguments, a short human-readable summary, structured machine-readable data, and extracted entities such as IDs and selected asset URLs. This format supports both conversational clients and applications that need to process the returned values programmatically.

Setup and configuration

Install the labeveryday/nba_mcp_server MCP server from PyPI with uvx nba-stats-mcp, or install the package with pip. The README also documents running it from a cloned source checkout with uv sync or an editable pip install. Claude Desktop can launch the server with uvx and the nba-stats-mcp package name, then requires a client restart after the configuration is changed.

Optional environment variables control logging and request behavior. NBA_MCP_LOG_LEVEL changes logging verbosity. Timeout, concurrency, retry, cache, and TLS verification settings are available through NBA_MCP_HTTP_TIMEOUT_SECONDS, NBA_MCP_MAX_CONCURRENCY, NBA_MCP_RETRIES, NBA_MCP_CACHE_TTL_SECONDS, NBA_MCP_LIVE_CACHE_TTL_SECONDS, and NBA_MCP_TLS_VERIFY.

Tools and capabilities

The labeveryday/nba_mcp_server MCP server provides tools for:

• Resolving player and team names to IDs.
• Listing games for a date, finding a matchup, and retrieving game details.
• Reading full box scores, play-by-play events, and player rotation data.
• Querying player biographies, season, career, game-log, hustle, defense, and advanced statistics.
• Retrieving awards, shooting splits, shot-chart data, rosters, schedules, standings, and league leaders.
• Comparing two players, summarizing a day’s games, and combining roster, record, and schedule information in a team overview.

Team advanced statistics include offensive rating, defensive rating, pace, and net rating. League leader queries support current-season, all-time, and hustle scopes, with categories such as points, assists, and rebounds.

Limitations and notes

The server depends on the availability and behavior of the official NBA APIs, so API responses and current-data coverage may vary. Live endpoints use a shorter default cache period than general statistics endpoints, while transient request failures are retried according to configuration.

The README documents public NBA CDN URLs for player headshots and team logos; these are included in selected responses and do not require an API key. The provided client setup specifically covers Claude Desktop, while Python/Strands usage is also shown. No credentials or required environment variables are specified in the setup instructions.