Sports Betting Mcp

io.github.seang1121由 GitHub 账号 seang1121 发布v1.0.0更新于 Oct 24, 2026

AI sports betting picks, odds, injuries & line movement. 1,353+ picks, 59.6% win rate.

已验证STDIO仅桌面金融数据与分析

概览

AI 生成的概览

让 AI 助手获取 NBA、NHL、NCAAB 和 MLB 的体育博彩推荐、实时赔率、伤病报告、盘口变动和模型表现数据。

可以做什么
把 MCP 客户端连接到体育博彩分析 API，提供约 12 个工具。工具涵盖当日推荐与最佳推荐、待结算与已结算推荐、单场比赛的 12 智能体分析、实时胜负盘/让分/总分赔率、盘口变动与伤病提醒、按联盟和投注类型划分的胜率、模型统计、用户排行榜以及系统健康检查。每次调用都返回结构化数据，助手可据此排序、筛选或解释。
适用场景
当你希望助手拉取这四个联盟的每日博彩推荐、赔率和市场信号，或在行动前核对已公布的推荐战绩时使用。适合已经关注体育博彩、并希望在自己的 AI 客户端内直接查看数据而非另开网站的用户。
运行要求
以 Python 包形式在本地运行（pip install sports-betting-mcp，需 Python 3.10+），通过 stdio 通信，因此需要本地 Python 运行环境。需要在 SPORTS_BETTING_API_KEY 环境变量中提供 API 密钥，密钥从提供方网站获取。可选变量 SPORTS_BETTING_API_URL 用于设置 API 基础地址，默认指向本机地址。需要能访问提供方的 API 网络。
安装前请注意
API 密钥属于凭据，通过 SPORTS_BETTING_API_KEY 环境变量传入，不要写入共享配置。该服务器提供的是博彩推荐与赔率，属于赌博相关信息，并非财务建议，据此下注可能损失金钱。部分接口可以记录推荐或注册 webhook，会向提供方服务写入数据。默认 API 地址指向本机，必须改为提供方主机才能正常工作。

安装

在 SourceWeft 中

  1. 打开 控制台中的 Sports Betting Mcp，将其添加到工作区。
  2. 为需要使用其工具的对话启用该服务。

Desktop only，通过 STDIO。 STDIO 服务会启动本地进程，因此需要 SourceWeft 桌面宿主。

其他 MCP 客户端

参照 仓库 中的启动说明。

README

sports-betting-mcp

The first MCP server for sports betting. Give any AI agent live access to picks, odds, game analysis, and performance tracking across NBA, NHL, NCAAB, and MLB.

[Status] [Python] [PyPI] [License] [MCP] [Sports]

mcp-name: io.github.seang1121/sports-betting-mcp

What It Does

Connects any MCP-compatible AI agent to a live sports betting analysis system. Every pick is logged before tip-off and resolved against final scores. Nothing is cherry-picked.

MetricValue
Sports CoveredNBA, NHL, NCAAB, MLB
Bet TypesMoneyline, Spread, Totals
Pick Source12-agent consensus model
Total Picks3,859+ resolved
Platform Users30
Tools12 MCP tools
AuthAPI key (X-API-Key header)

Results by Sport

SportPicksWinsWin Rate
NBA1,26776260.1%
NHL1,14865657.1%
NCAAB1,14954947.8%
MLB28310938.5% (launched Apr 2026)

All results are queryable in real-time via the get_win_rate tool. Numbers update after every game.

Quick Start

Install

bash
pip install sports-betting-mcp

Add to Your MCP Client

Drop this into your MCP config (Claude Desktop, Cursor, Windsurf, Claude Code, etc.):

json
{  "mcpServers": {    "sports-betting": {      "command": "sports-betting-mcp",      "env": {        "SPORTS_BETTING_API_URL": "https://sportsbettingaianalyzer.com",        "SPORTS_BETTING_API_KEY": "your_api_key"      }    }  }}

Environment Variables

VariableRequiredDescription
SPORTS_BETTING_API_KEYYesYour API key for authentication
SPORTS_BETTING_API_URLNoAPI base URL (defaults to http://localhost:5000)

Works With

Any client that supports the Model Context Protocol:

ClientStatus
Claude DesktopFully supported
CursorFully supported
WindsurfFully supported
Claude Code (CLI)Fully supported
Any MCP ClientFully supported via stdio transport

Available Tools

12 tools. Every call returns structured data that AI agents can reason over, display, or act on.

Picks & Analysis

ToolWhat It Does
get_todays_picksAll AI picks with confidence scores and edge breakdowns, filterable by sport
get_top_pickSingle highest-confidence pick of the day
get_pending_picksCurrently unresolved picks that are still in play
get_completed_picksRecently resolved picks with W/L results -- verify the track record
analyze_gameFull 12-agent analysis on any game: consensus pick + edge breakdown

Odds & Market Data

ToolWhat It Does
get_live_oddsLive moneyline, spread, and totals for today's games
get_alertsActive alerts from the multi-agent system (line moves, injury impacts)

Performance & Stats

ToolWhat It Does
get_win_rateWin rate with breakdown by sport and bet type
get_model_statsModel performance: total picks, last-20 win rate, confidence tier breakdown
get_leaderboardUser rankings by win rate
get_system_statusHealth check -- uptime, database, scheduler status

How the Analysis Works

Each game runs through a multi-agent pipeline:

  1. 12 specialized agents evaluate the game independently -- covering momentum, matchups, injuries, rest, travel, public betting percentages, sharp money indicators, historical trends, and more.
  2. A consensus engine synthesizes all 12 opinions into a single pick with a confidence score.
  3. Edge calculation compares the model's implied probability against the current market line.
  4. Picks are logged before tip-off and resolved against final scores. No retroactive edits.

The confidence score and edge breakdown are included in every pick response, so your AI agent can filter, rank, or explain the reasoning behind any recommendation.

Full API — 22 Endpoints

The API has two tiers: Free (any API key) and Pro (approved access).

Free Tier (available to all API key holders)

EndpointMethodWhat It Does
/xk/pGETToday's AI picks with confidence scores
/xk/o/{sport}GETLive odds from FanDuel/BetMGM
/xk/wGETWin rate breakdown by sport and bet type
/xk/iGETActive injury report from Covers.com
/xk/mGETSignificant line movement (sharp money signals)
/xk/gamesGETToday's full schedule with odds snapshot and AI pick flags
/xk/statsGETModel performance by sport, bet type, confidence tier
/xk/lbGETLeaderboard — AI vs human win rates
/xk/newsGETSports news feed from RSS sources
/xk/qGETCurrently pending (unresolved) picks

Pro Tier (requires approved access)

EndpointMethodWhat It Does
/xk/fullGETFull structured picks — complete edge breakdowns, positive/negative edges, reasoning, scorecard, opposing side with odds, line movement, quality flags
/xk/analyzePOSTOn-demand 12-agent analysis for any game — send team names, get full consensus
/xk/resultsGETResolved picks with W/L, actual scores, profit, top contributing edges
/xk/clvGETClosing Line Value analysis — did the line move in our favor?
/xk/searchGETSearch all picks by team name with W/L record
/xk/logPOSTLog a pick (supports flip flag, pick source tracking)
/xk/log/batchPOSTBatch log up to 20 picks with per-pick success/duplicate/error status
/xk/historyGETYour pick history — pending, resolved, win rate
/xk/slipPOSTGenerate Nimrod bet slip image
/xk/webhookPOST/GET/DELETERegister webhooks for pick events
/xk/fullGETIncludes fade/flip data for every pick — opposing side, line, odds

Pro Pick Payload

Every pick from /xk/full includes:

json
{  "ai_pick": "Lakers -5.5",  "ai_verdict": "STRONG BET",  "ai_probability": "65%",  "ai_reasoning": "Lakers defense ranks top 5, opponent on B2B...",  "edges": [13 individual edge factors],  "top_edges": [top 5 positive],  "negative_edges": [top 3 negative],  "scorecard": [3 model scores],  "fade_pick": "Celtics +5.5",  "fade_team": "Celtics",  "fade_odds": "+105",  "fade_note": "Betting AGAINST the AI — take Celtics +5.5",  "line_moved": true,  "flags": ["line_moved"],  "data_quality": "good"}

Pick Source Tracking (for learning)

When logging picks, include pick_source to track decision types:

  • model_agree — following the AI's pick
  • flip — fading/betting against the AI
  • manual_override — custom pick

This enables win rate comparison between model-agree vs flip picks over time.

API Authentication

Authenticate with X-API-Key header. Get a free key at sportsbettingaianalyzer.com. Pro access requires admin approval.

Tech Stack

ComponentTechnology
RuntimePython 3.10+
ProtocolMCP (Model Context Protocol)
Transportstdio
BuildHatchling
DistributionPyPI (sports-betting-mcp)
BackendFlask + SQLite
Analysis12-agent consensus pipeline
SportsNBA, NHL, NCAAB, MLB

Related Projects

License

MIT

来源：README.md，提交 e0fa825

工具

0
SourceWeft 尚未在沙箱中启动过该软件包，因此其工具未知。

版本历史

1
  1. v1.0.0最新Sep 16, 2026

评分

已满足适用检查项的 50%

该评分衡量此收录条目的完整程度，依据注册中心条目、端点与公开来源计算得出。它不是对服务器本身的评价，也不表示服务器安全或当前可用：可连通性是带日期的检查结果，评分也不包含安装量或用户评价。

  • 近 30 天内可连通25 分

    请保持端点能够响应 MCP 握手；SourceWeft 每 30 天探测一次。

  • 已声明认证方式10 分

    请在注册中心条目中声明凭据请求头，或保持端点能够响应探测，以便判定其认证方式。

  • 已列出工具15 分

    请让服务器在无凭据时响应 tools/list，或发布可在沙箱中无凭据启动的软件包。

  • 描述必需15 分
  • 图标10 分
  • 文档15 分
  • 许可证10 分