Polymarket Predictions MCP is an MCP (Model Context Protocol) server used to provide real-time market odds from Polymarket, enabling AI agents and analysts to access, compare, and act on decentralized prediction data.
Purpose: This tool is designed to connect Polymarket's prediction market data into AI workflows, making it convenient for users to query events, market information, and perform searches and analysis.
Main capabilities:
- Event retrieval (get_events): Retrieve a list of Polymarket events and their associated markets, outputting detailed information including titles, descriptions, end dates, trading volume, etc., and displaying market questions, status, trading volume, outcomes, and outcome prices in a table. Supports sorting by fields such as trading volume, pagination, filtering by ID, slug, and tags, as well as filtering closed events.
- Market retrieval (get_markets): Retrieve a list of Polymarket markets, outputting a table containing questions, options (outcome and price combinations), end dates, trading volume, and closed status. Supports pagination, sorting, filtering by ID, slug, and condition ID, as well as filtering closed markets.
- Event search (search_events): Use Polymarket's /public-search endpoint to search events, supporting parameters such as query terms, caching, event status, number of results per type, pagination, event tags, number of closed markets to retain, sort field, ascending order, search tags, search profiles, duplicate type, excluded tag IDs, and optimized search. The output format is similar to get_events.
- Prompt support (analyze_market): Provide a prompt template for analyzing a specific market (by slug), encouraging the use of tools to obtain details and provide insights about probability, trading volume, and potential outcomes.
Usage: Environment requirements: Python 3.10 or higher; uv is recommended for dependency management, and pip can also be used. Installation steps:
- Clone the repository: git clone https://github.com/kukapay/polymarket-predictions-mcp.git cd polymarket-predictions-mcp
- Install dependencies: Using uv: uv sync Or using pip: pip install mcp[cli] httpx tabulate
- Install to Claude Desktop: uv run mcp install main.py --name "Polymarket Predictions" Configuration file reference: { "mcpServers": { "Polymarket Predictions": { "command": "uv", "args": [ "--directory", "/path/to/polymarket-predictions-mcp", "run", "main.py" ] } } } Replace /path/to/polymarket-predictions-mcp with the actual installation path.
The server exposes three tools and one prompt: get_events, get_markets, search_events, and analyze_market.
Applicable scenarios: Suitable for AI applications that require real-time prediction market data, such as financial analysis, event forecasting, market monitoring, etc.
Limitations: This tool currently relies on Polymarket's public API; specific usage limits (such as rate limits) are not provided at this time. The project uses the MIT license.
