Pull this week’s NFL prop board and write it to a CSV with the best venue per prop.
What you get back: The live board as structured rows your agent can save, chart or diff against yesterday.
Kalshi and Polymarket data in Cursor and VS Code
Quick Answer
to give Cursor or VS Code (GitHub Copilot agent mode) live Kalshi and Polymarket data, click a one-click install button below, or add https://predictionmarketspicks.com/api/mcp/mcp as an "http" MCP server in your MCP config. No API key: 22 of the 33 tools answer immediately.
MCP server URL
https://predictionmarketspicks.com/api/mcp/mcpSet it up by hand
- Cursor: open Settings → MCP (or edit ~/.cursor/mcp.json) and add the block below under mcpServers.
- VS Code: the same entry goes under "servers" in .vscode/mcp.json, then start it from the MCP view. GitHub Copilot uses it in agent mode.
- Save, confirm predictionmarketspicks shows as running, and ask the agent about a market.
mcpServers JSON (Cursor)
{
"mcpServers": {
"predictionmarketspicks": {
"type": "http",
"url": "https://predictionmarketspicks.com/api/mcp/mcp"
}
}
}Five prompts for a coding agent
Copy one into a new chat and swap in your own team, player or market.
Scan for Kalshi vs Polymarket price gaps on NBA, NHL and MLB games and print them as a table.
What you get back: The largest gap in percentage points with the cheaper venue and its WATCH or ARB signal.
Show the Kalshi 15-minute boards and write a script that alerts me when a window moves 10¢.
What you get back: The current windows with prices and close times — the shape a polling script needs.
Give me the Fed odds for the next meeting as JSON.
What you get back: The next meeting’s strike breakdown on Kalshi with a Polymarket and CME check, ready to serialise.
Compute the EV of buying YES at 37¢ if my model says 44%.
What you get back: The expected-value edge in percent and a BUY, SELL or SKIP read.
Cursor, VS Code and prediction-market data — frequently asked
Is this an alternative to the Kalshi API for developers?
For reading prices, often yes: the tools return Kalshi and Polymarket prices, ladders and boards with no key to manage. To place orders you still need the exchange’s own API and account — these tools are read-only.
Does the server return structured data my code can use?
Yes. The board and list tools return structured content alongside their text answer, so an agent can write rows to a file, chart them or compare runs without parsing prose.
Does it work with GitHub Copilot?
Yes, in VS Code agent mode. Add the server with the one-click button or under "servers" in .vscode/mcp.json, start it, and Copilot can call the tools.
What are the usage limits?
22 tools are free with depth caps on the list-shaped ones; deeper results unlock instantly from the result itself. Pro adds the model’s edge calls and live mispricing scans.
Using ChatGPT, Claude, Gemini or Grok? Same URL — setup for every host → · what every tool returns →