Open-source trading engine and MCP server for prediction market agents.
Connect AI agents to Kalshi and Polymarket via 13 MCP tools and Python scripts.
Oracle3 is an open-source trading engine and Model Context Protocol server that connects AI agents to prediction markets like Kalshi and Polymarket. It is built for quantitative developers and researchers who want to automate market discovery, fee-aware arbitrage checks, and order execution.
The engine discovers contracts, inspects order books, and computes fee-adjusted net edge across related event constraints like implications and exclusions. It provides 13 MCP tools, six reusable agent skills, and Python interfaces for paper trading or live execution on Kalshi, Polymarket, and Solana.
It suits quantitative developers, algorithmic traders, and AI researchers building autonomous market agents. It does not suit casual prediction market bettors looking for a simple web interface or users without Python development experience.
Provides 13 structured tools for AI clients to query market data, quotes, and order books.
Calculates net edge across five related contract constraints including implications and event sums.
Includes six built-in workflows for data discovery, constraint arbitrage, and paper operations.
Supports paper trading and live execution adapters for Kalshi, Polymarket, and Solana via DFlow.
Free under its open-source license. The only cost is running it yourself, plus any capital risked in live trading.
877 open remote AI-training roles, updated November 16, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.
Oracle3 supports Kalshi and Polymarket natively, alongside Solana via DFlow for live execution.
Public Kalshi and Polymarket market reads require no exchange credentials; keys are only needed for live trading.
The engine includes constraint checks that compute per-leg fees and net edge against default fee schedules.
See Zekai first in Google