# puntersedge [Health: Active]

**Category:** 💻 Developer Tools  
**Repository:** https://github.com/Propertyscout001/puntersedge-mcp  
**GitHub Stars:** 0  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/puntersedge

## Description
Australian and New Zealand racing plus Australian sports odds for AI agents. Read-only tools.

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

```json
"mcpServers": {
  "puntersedge": {
    "command": "uvx",
    "args": ["puntersedge-mcp"],
    "env": {
      "PUNTERSEDGE_API_KEY": ""
    }
  }
}
```

**Requires environment variables:** `PUNTERSEDGE_API_KEY` — the values above are empty placeholders; fill in real credentials before running (see the repository for what each one is for).

## Documentation

## What puntersedge MCP server does

The puntersedge MCP server exposes a read-only PuntersEdge odds feed to MCP-compatible AI clients. It covers Australian thoroughbred, greyhound, and harness racing, New Zealand thoroughbred and harness racing, and sports markets including AFL, NRL, NBA, NFL, tennis, and cricket. Sports coverage can change, so `list_sports` shows the currently available sports and their keys.

Racing tools return schedules, runner prices, market movements, and settled results. Racing data can include prices from multiple bookmakers, runner details, scratchings, track conditions, weather, data age, and stale-data indicators. Sports tools return bookmaker odds for a selected sport and requested market types. All nine tools are read-only: they do not place, modify, cancel, or settle bets.

## How it works

The server runs as a standard stdio MCP process and sends requests to the PuntersEdge API. It reads `PUNTERSEDGE_API_KEY` from the environment at startup and does not return the key in tool results. The TypeScript repository is a smaller implementation than the separate Python package; this repository contains five racing tools and four sports or account tools.

The main racing query is `racing_next_to_go`, which returns imminent Australian races and bookmaker prices. Use `racing_events` when a schedule without prices is sufficient, and `racing_best_odds` when only the highest available price per runner is needed. `racing_movers` identifies notable market firming or drifting, while `racing_results` returns completed-race outcomes and dividends.

Sports requests use `get_sports_odds`, `get_best_odds`, and `list_sports`. Do not pass `horse-racing` as a sports key; racing is handled by the separate `racing_*` tools. API credit headers are included in tool results, and `account_usage` reports monthly usage without consuming credits.

## Setup and configuration

Run the repository directly with:

```bash
npx -y github:Propertyscout001/puntersedge-mcp
```

The first run clones and compiles the project, so Git, Node.js 18 or newer, and a working Node toolchain are required. Obtain a PuntersEdge API key and provide it through the environment:

```json
{
  "mcpServers": {
    "puntersedge": {
      "command": "npx",
      "args": ["-y", "github:Propertyscout001/puntersedge-mcp"],
      "env": { "PUNTERSEDGE_API_KEY": "your_key_here" }
    }
  }
}
```

The same command, arguments, and environment variable can be used by other hosts that accept stdio MCP servers. The README provides configuration locations and examples for Claude Desktop, Claude Code, Cursor, Windsurf, and VS Code.

## Tools and capabilities

- `racing_next_to_go`: upcoming Australian races with bookmaker prices.
- `racing_best_odds`: best listed price for each runner and its bookmaker.
- `racing_events`: upcoming venues, race numbers, and jump times.
- `racing_results`: settled results and dividends.
- `racing_movers`: runners showing significant market movement.
- `list_sports`: currently supported sports and sport keys.
- `get_sports_odds`: odds for a sport and selected markets.
- `get_best_odds`: best price for each sports selection.
- `account_usage`: credits used and remaining for the month.

API calls consume credits according to the underlying endpoint. The free tier provides 3,000 credits per month, and exhausted allowances stop requests rather than creating overage billing.

## Limitations and notes

The puntersedge MCP server reads published prices but does not predict results, recommend bets, or access bookmaker accounts. It has no bookmaker login, session, or wagering capability. Account administration such as key rotation, billing, and webhook registration is outside this server.

Prices may be stale if an upstream feed has stopped updating. Responses expose `age_seconds` and a `stale` flag, and can identify stale bookmakers. Treat prices supported by only one bookmaker cautiously. The server also depends on the PuntersEdge API key and available monthly credits; a `402` response indicates that the allowance has been used.

_Full upstream README: https://allmcps.com/mcp/puntersedge/readme_

