Turn Polymarket into an AI-ready prediction feed.
This site is the front door for the open-source polymarket-mcp MCP server. It lets AI agents access Polymarket markets, prices, and history through a clean MCP interface.
Example agent: Odds Sentinel
What is polymarket-mcp?
polymarket-mcp is a Model Context Protocol (MCP) server that wraps Polymarket's APIs. It standardizes several tools for AI agents to interact with prediction markets in a consistent way.
The server provides standardized tools for:
- Listing markets with filtering and pagination
- Getting detailed market information
- Fetching current prices and implied probabilities
- Retrieving historical price and volume data
It's designed to run locally or on your own infrastructure, and integrates seamlessly with MCP-compatible AI clients like Claude Desktop.
Highlights
- ✓Clean JSON responses optimized for LLMs
- ✓Access to real prediction markets: politics, crypto, sports, and more
- ✓Historical data ready for charting and backtests
- ✓Error handling and rate limits baked into the server
In Simple Terms
Think of polymarket-mcp as a translator between AI assistants (like Claude) and Polymarket prediction markets.
What are prediction markets? They're places where people bet on whether something will happen—like "Will it rain tomorrow?" or "Who will win the election?" The prices tell you how likely people think each outcome is.
What does this tool do? It lets AI assistants ask questions about these markets and get easy-to-understand answers. Instead of you having to visit Polymarket's website and search through data, you can just ask your AI assistant: "What are the current odds on the election?" or "Show me the most popular markets right now."
Why is this useful? AI assistants can help you understand market trends, track changes over time, and even alert you when something interesting happens—all through a simple conversation, without needing to be a tech expert.
Quickstart: from zero to running MCP server
Clone, configure, and run the polymarket-mcp server in three simple steps.
1. Clone & install
Get the repository and install dependencies using uv.
git clone https://github.com/polysolmcp/polysolmcp.git
cd polysolmcp
uv pip install -e .2. Set your environment variables
Configure your Polymarket API key and wallet address.
KEY=your_polymarket_api_key_here
FUNDER=your_polymarket_wallet_addressCheck the GitHub README for the latest required variables and how to obtain your API key.
3. Run the server
Start the MCP server and optionally use the inspector for debugging.
uv run src/polymarket_mcp/server.pynpx @modelcontextprotocol/inspector \
uv --directory /path/to/polysolmcp \
run src/polymarket_mcp/server.pyThe inspector is optional, but great for debugging MCP tools.
Connect to Claude Desktop (example config)
Claude Desktop can be configured to talk to MCP servers via its claude_desktop_config.json. Add polymarket-mcp under the mcpServers section.
{
"mcpServers": {
"polymarket-mcp": {
"command": "uv",
"args": [
"--directory",
"/Users/your-username/path/to/polysolmcp",
"run",
"src/polymarket_mcp/server.py"
],
"env": {
"KEY": "your_polymarket_api_key",
"FUNDER": "your_polymarket_wallet_address"
}
}
}
}macOS path:
~/Library/Application Support/Claude/claude_desktop_config.jsonWindows path:
%APPDATA%/Claude/claude_desktop_config.jsonPaths can vary, and you should consult Claude Desktop documentation if needed.
Tools your agents can call
The polymarket-mcp server exposes four main tools to your AI agents.
list-markets
Browse markets with filters for status and pagination.
get-market-info
Fetch detailed info for a specific market.
get-market-prices
Retrieve current outcome prices and implied probabilities.
get-market-history
Get historical prices and volumes for a market.
Agent blueprints you can build
A few ready-made concepts to get your creativity going.
Market Explorer Bot
Scans open markets, surfaces high-liquidity, high-volume markets, grouped by category.
Tools used:
list-markets, get-market-info
Example instruction:
"Regularly scan open Polymarket markets, rank them by liquidity and 24h volume, and produce a concise summary of the top 10, grouped by category (politics, crypto, sports, etc.). Highlight any markets that have recently opened or significantly changed volume."
Odds Change Sentinel
Watches specific markets for sharp moves in probabilities and alerts the user.
Tools used:
get-market-prices, get-market-history
Example instruction:
"Monitor the specified Polymarket market IDs, track their price history over the last 24 hours, and alert me whenever the implied probability moves by more than 5 percentage points within an hour. Provide context from the recent history when you generate an alert."
Narrative Tracker
Follows a theme (e.g. US election, BTC ETF, etc.) across multiple related markets.
Tools used:
list-markets, get-market-info, get-market-prices
Example instruction:
"Identify markets related to the US 2024 election, group them by candidate or key event, and periodically summarize how the probabilities have shifted across the narrative. Point out any markets where the odds diverge significantly from the others."
FAQ
Do I need to trust this website with my API key?
No. The polymarket-mcp server runs on your own machine or infrastructure. This site is just a web interface for connecting and testing; we never see your keys.
Which AI clients support MCP?
Claude Desktop explicitly supports MCP, along with any other MCP-compatible clients. The protocol is open and standardized.
Can I deploy the MCP server to my own infrastructure?
Yes, it's standard Python. You can run it anywhere you can run uv and provide environment variables—your local machine, a VPS, a container, or any other infrastructure you control.
Does this perform trading on my behalf?
No. The MCP server provides data about markets, prices, and history. Any actual trading, wallet actions, or automation is done by your own agents or scripts and is your responsibility.