prediction-market

This MCP server exposes live prediction market data from Polymarket, PredictIt, and Kalshi for use in AI workflows. It does not require API keys and provides current odds or prices across multiple platforms in a single interface. The server includes a built-in tool called get-prediction-markets that lets you search across platforms using a keyword, returning matching markets with platform-specific formats (percentage odds for Polymarket, dollar prices for PredictIt, and regulated US markets for Kalshi). Typical use cases include feeding real-time market data into AI prompts, agents, or notebook workflows to drive decision-making or research analyses. You can invoke the MCP tool via standard MCP client tooling or through Cursor integrations that consume the server’s model-context data.

Usage examples include querying for markets by keyword (e.g., trump, election) to retrieve current market data across platforms, or integrating the tool into automated workflows that monitor and react to market odds and prices in real time.

Prerequisites:

• Node.js 18+ and npm installed on your machine
• Access to the internet to install the MCP package

Manual installation steps:

1. Clone the repository or install via npm globally npm install -g prediction-markets-mcp

2. Build (if needed by the project distribution) git clone https://github.com/JamesANZ/prediction-markets-mcp.git cd prediction-markets-mcp npm install npm run build

3. Run the server locally node build/index.js

4. If you prefer to run via npx (no local install required) npx -y prediction-markets-mcp

Notes:

• The server is built with Node.js and TypeScript. If you clone from source, you’ll typically run npm install followed by npm run build to generate the runnable bundle before starting the server.

Tips and considerations:

• The MCP supports multi-platform data, so you’ll see odds/prices aggregated from Polymarket, PredictIt, and Kalshi in a unified interface.
• No API keys are required for basic data access, but data availability depends on the respective platforms’ public endpoints.
• The recommended quick-start for Cursor users is to use the one-click install link, which encodes the proper MCP config (npx prediction-markets-mcp).
• If you encounter rate limits or platform-specific data delays, consider building and caching results in your workflow to smooth out bursts.
• Ensure Node.js version compatibility (18+) when running from source builds.