mcp-cpbl-statistics 0.1.0

MCP server for CPBL statistics

MCP CPBL Statistics

繁體中文

An MCP server for CPBL (Chinese Professional Baseball League / 中華職棒) statistics, exposing AI-callable tools over Model Context Protocol (MCP).

What This Does

• 10 ready-to-use tools covering team standings, player profiles, batting/pitching stats, game logs, head-to-head matchups, top-5 leaderboards, and game schedules
• MCP server (stdio JSON-RPC 2.0) — plug into Claude Code and start asking about CPBL instantly
• Web scraper — HTML scraping and XHR JSON API calls against cpbl.com.tw
• No-auth public access — No API keys required; CSRF token handling is automatic
• Fixture-based unit tests — Fast offline test suite with saved HTML/JSON fixtures
• Integration tests — Live end-to-end tests against cpbl.com.tw, opt-in via -m integration

---

Quick Start

Install

pip install mcp-cpbl-statistics

Or with uv:

uv sync

Use with Claude Code

claude mcp add --transport stdio cpbl-statistics -- mcp-cpbl-statistics

Use with Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "cpbl-statistics": {
      "command": "uv",
      "args": ["run", "mcp-cpbl-statistics"]
    }
  }
}

Or with pip:

{
  "mcpServers": {
    "cpbl-statistics": {
      "command": "mcp-cpbl-statistics"
    }
  }
}

---

Tools (10)

Standings

Tool	Description
get_season_standings	Current season team standings (W/L record, pitching, batting, fielding)
get_history_standings	Historical standings for a given year (first half / second half / full season)

Top Lists

Tool	Description
get_toplist	Season top-5 leaderboards for ERA, W, SV, HLD, SO, AVG, H, HR, RBI, SB

Players

Tool	Description
search_players	Search active players by name (partial match); returns player ID, name, team
get_player_profile	Player profile: jersey number, position, batting/throwing hand, height/weight, birthday, first appearance, education, draft
get_player_stats	Career or single-season batting/pitching stats for a player
get_player_apart_stats	Split stats (home/away, vs. opponent, lineup position, runners on base, inning, score situation, month, stadium, etc.)
get_player_game_log	Per-game log for a player, optionally filtered by year or last N games
get_player_headtohead	Head-to-head matchup stats for a player against a specific team

Schedule

Tool	Description
get_schedule	Game schedule / results, filterable by year, month, team, and game type; finished games include score, W/L/SV pitchers, and MVP

kind_code values (shared across tools)

Code	Description
A	一軍例行賽 (1st team regular season, default)
B	一軍明星賽 (All-Star game)
C	一軍總冠軍賽 (Championship series)
D	二軍例行賽 (2nd team)
E	一軍季後挑戰賽 (Postseason challenger)
G	一軍熱身賽 (Spring training)

---

Project Structure

mcp-cpbl-statistics/
├── main.py                              # Stub entry point (unused)
├── pyproject.toml                       # Project metadata & scripts
├── .mcp.json                            # Local MCP server config
│
└── src/mcp_cpbl_statistics/
    ├── server.py                        # FastMCP server, registers all tools
    ├── scraper/
    │   └── fetcher.py                   # HTTP client (HTML scraping + JSON POST APIs)
    ├── models/                          # Pydantic output models
    │   ├── standings.py
    │   ├── toplist.py
    │   ├── player.py
    │   ├── apart.py
    │   ├── game_log.py
    │   ├── headtohead.py
    │   ├── schedule.py
    │   └── teams.py                     # Team code registry
    └── tools/                           # MCP tool implementations
        ├── standings/                   # get_season_standings, get_history_standings
        ├── toplist/                     # get_toplist
        ├── player/                      # search_players, get_player_*, ...
        └── schedule/                    # get_schedule

Testing

Unit tests run against saved HTML/JSON fixtures (no live network calls required):

uv run pytest

Integration tests make live requests to cpbl.com.tw and are excluded by default:

uv run pytest -m integration

Data Source

All data is scraped from the official CPBL website (cpbl.com.tw). No API keys are required. The scraper handles CSRF token extraction automatically for POST-based JSON APIs.

License

MIT