Documentation
Scanna
The agent-native intelligence layer for prediction markets — real-time signals, ML mispricing detection, smart money tracking, and cross-venue arbitrage across Polymarket and Kalshi, delivered through an API, an MCP server, SDKs, and more.
The agent-native intelligence layer for prediction markets. Scanna transforms real-time off-chain data and deep on-chain analytics into high-fidelity signals — delivering real-time insights to forecast events, identify pricing inefficiencies, and capture structural alpha with speed and accuracy.
20,000+ markets · 20.5M trades · 407K wallets
Products
REST API
The core intelligence API — heat scores, ML predictions, smart money, divergence, and raw market data over HTTP. Live.
MCP Server
Scanna as tools inside Claude, Cursor, and other AI agents — connect in one command. Live (beta).
Marketplace
The Scanna plugin for Claude Code — skills, slash commands, and the MCP server. Beta.
SDKs
Typed TypeScript and Python clients over the API. Pre-release.
Simulation
A proprietary multi-agent simulation engine for scenario modeling. Private beta.
Mobile · Scanna Pulse
A consumer prediction-market chatbot for iOS and Android. Private beta.
Get started
Request access
Get an API key or use the Flex tier to pay per request via x402. One key works across the API, MCP, and SDKs.
Quickstart
Make your first authenticated request in minutes.
What the intelligence gives you
- Trending signals — a ranked feed of the hottest markets by a composite heat score, with whale activity and human-readable signals.
- ML predictions — independent probability estimates for resolved and live markets from a daily-retrained model. See AI Predictions.
- Smart money — performance-ranked wallets so you can see where informed capital is moving. See Smart Money.
- Cross-venue arbitrage — price gaps for the same event across Polymarket and Kalshi via Divergence.
- Raw data — markets, prices, order books, trades, holders, wallets, and leaderboards. See Raw Data.
- Training data — labeled, time-sliced features for your own models via Training Data Export.