Question 1

What is Orrery?

Accepted Answer

Orrery is a public, read-only verification layer for prediction markets — primarily Polymarket. It aggregates and explains market movement, source context, resolution risk, large-trade flow, wallet activity, and the Daily Brief. Orrery is not affiliated with Polymarket and does not facilitate trading.

Question 2

How can an AI agent use Orrery?

Accepted Answer

Three integration surfaces: (1) free public read-only API at /api with OpenAPI spec at /openapi.yaml; (2) MCP server at /api/mcp/v1 for Claude Desktop / Cursor / Claude Code (one config line, tools/resources/prompts surfaced); (3) paid x402 endpoints at /api/x402/v1 for autonomous agents settling in USDC on Base. Every page also has a machine-readable Markdown variant at /<route>/markdown for citation. Discovery manifest at /.well-known/ai-plugin.json.

Question 3

Does Orrery predict outcomes or recommend trades?

Accepted Answer

No. Orrery describes — never predicts. Every signal carries explicit Evidence (live confidence) and Backtest (historical performance) tiers, and an Action recommendation that is always one of: Watch only / Inspect timeline / Create alert / Verify source / Ignore — too noisy. Never Buy or Sell. AI explanations are grounded in supplied data and are required to refuse trade-recommendation framing.

Question 4

How fresh is the data?

Accepted Answer

Live. Per-market terminal pages revalidate every 30 seconds; the home dashboard every 60 seconds; the Daily Brief every 5 minutes. Each block carries an 'updated Xs ago · source: …' timestamp. Orrery participates in IndexNow so search engines and grounding pipelines re-index updated pages within seconds.

Question 5

How does Orrery determine the resolution source for a market?

Accepted Answer

Polymarket frequently leaves the resolutionSource field empty and inlines the source in the description ('resolves according to Binance BTC/USDT close'). Orrery runs a deterministic regex extractor that tags the source with a type (exchange_price / official_government / sports_official_result / court_record / company_filing / news_consensus / social_media_post / ambiguous) and a confidence (low / medium / high). The result drives the Resolution & Risk block on per-market pages and the resolution_risk signal kind.

Market	Score	Breakdown (Evidence / Liquidity / Spread / Res. Risk)	Next Step
Will no Fed rate cuts happen in 2026?	84	81/87 90/80	Inspect timeline
Roland Garros WTA: Maja Chwalinska vs Mirra Andreeva	72	53/91 80/80	Compare siblings

Orrery — the verification layer for prediction markets

5 things to verify