# AI Ball — AI football match analysis [Health: Active]

**Category:** 💻 Developer Tools  
**Repository:** https://github.com/aiballfooty/aiball-mcp  
**GitHub Stars:** 0  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/ai-ball-ai-football-match-analysis

## Description
Football match outcome probabilities recorded before kick-off, and an open record of results.

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "ai-ball-ai-football-match-analysis": {
    "command": "npx",
    "args": ["-y","github:aiballfooty/aiball-mcp"]
  }
}
```

## Documentation

## What AI Ball — AI football match analysis does

AI Ball — AI football match analysis gives MCP-compatible assistants access to football forecasts and their subsequent results. The data covers home, draw, and away probabilities captured before kick-off, the model’s most likely outcome, confidence information, and the final result once a match has been played.

The record is intended to remain open: successful and unsuccessful predictions are both counted. This makes the server suitable for checking individual forecasts as well as comparing performance over time.

## How it works

The local server calls AI Ball’s public record API and returns responses assembled from a defined set of fields. It does not require authentication. Requests include tool parameters such as the requested date, time zone, and language, along with a versioned user-agent. Conversation content, local files, and user identifiers are not sent to the API.

Responses are cached for five minutes, and requests are spaced to avoid repeatedly querying the upstream service. Match listings distinguish between upcoming, pending, and finished matches. Upcoming predictions are locked before kick-off, while pending readings may still change.

The balanced model stores home, draw, and away fractions that total one. A result is marked as a hit when the model’s highest-probability outcome occurred. Record statistics also include a favourite-pick baseline and a one-in-three random baseline. Percentages can be null when a confidence band has fewer than the required minimum sample; the accompanying hit and total counts remain available.

## Setup and configuration

For a local installation, use Node.js 18 or newer. Claude Desktop can load the downloadable `aiball-mcp.mcpb` package from the latest release. Claude Code can register the server with `npx -y github:aiballfooty/aiball-mcp`, and Cursor or VS Code can use an MCP configuration with `npx` as the command and the GitHub repository as its argument.

A hosted Streamable HTTP endpoint is available at `https://aiball.samagent.ai/mcp`, requiring no sign-in. The hosted service is read-only and limits clients to 60 requests per minute per IP. Set `AIBALL_API_BASE` to use a different API base URL when running the local server.

## Tools and capabilities

The local server provides three tools:

- `list_matches` returns matches for one calendar day, including identifiers, kick-off times, leagues, teams, and status.
- `get_match_analysis` returns one match’s recorded probabilities, most likely outcome, confidence, capture time, and—after the match—its score and hit status.
- `get_open_record` returns the complete record or a specified week, including match totals, successful predictions, comparison baselines, and confidence-band counts.

Tools support English and Malay names through the `lang` parameter. The hosted endpoint also provides `get_match_context`, which supplies head-to-head, recent-form, injury, and fixture information for finished matches.

## Limitations and notes

AI Ball — AI football match analysis is read-only; it cannot create, modify, or remove records. The local tools access only the public record API, and no account or API key is supported or needed. Pending-match readings may change before completion, so they should not be treated as final forecasts. The service is provided for information only and is not advice.

_Full upstream README: https://allmcps.com/mcp/ai-ball-ai-football-match-analysis/readme_