What is MCP?
The Model Context Protocol (MCP) is an open standard that lets AI agents call external tools in a structured way. Instead of an agent scraping a website or parsing unstructured text, it calls typed functions with parameters and receives structured responses.
SimpleFunctions as an MCP Server
SimpleFunctions runs as an MCP server that exposes tools like:
scan_markets: Find markets matching criteria with live pricesget_edges: Get current edge calculations for a thesisevaluate_thesis: Trigger a thesis evaluation with new informationget_orderbook: Get live orderbook depth for a specific market
Why This Matters
Any AI agent that supports MCP can now:
- Research prediction markets by calling
scan_markets - Build and evaluate theses by calling
evaluate_thesis - Find trading opportunities by calling
get_edges - Monitor positions by calling the relevant tools
This turns prediction market analysis from a human-only activity into something any agent can do.
Setting Up the MCP Server
The SimpleFunctions CLI includes a built-in MCP server:
sf mcp serve
This starts a local MCP server that Claude Desktop, Cursor, or any MCP-compatible agent can connect to. The server handles authentication, rate limiting, and data formatting automatically.
Common Use Case
Connect SimpleFunctions MCP server to Claude Desktop. Then ask Claude: "What prediction markets have the biggest edge on recession right now?" Claude calls scan_markets and get_edges under the hood, returning a structured analysis.