# Foresea

> Foresea turns prediction-market questions into calibrated probability forecasts
> with supporting evidence, a written rationale, and the model-vs-market edge.
> Free to use, with an open JSON API agents can call directly.

## Use the API
- [Agent integration guide](https://foresea.ink/agents): practical entry point for AI
  agents, with MCP, OpenAPI, streaming, radar, and track-record examples.
- [Agent manifest](https://foresea.ink/.well-known/agent.json): machine-readable
  endpoints, recommended workflow, and resource list.
- [Remote MCP server](https://foresea.ink/mcp/): Streamable HTTP MCP endpoint for agents.
  Tools: `foresea_forecast`, `foresea_analyze_market`, `foresea_scan_markets`,
  `foresea_radar`, `foresea_track_record`, `foresea_pr_agent`. Discovery manifest:
  `https://foresea.ink/.well-known/mcp/server.json`.
- [Forecast](https://foresea.ink/docs): `POST https://foresea.ink/predict` with `{"question": "..."}` returns a
  structured forecast (binary / multiple-choice / numeric / date), a confidence,
  a rationale, and relevant evidence sources. No auth required.
- [Streaming forecast](https://foresea.ink/docs): `POST https://foresea.ink/predict/stream`
  returns SSE events (`meta`, `delta`, `done`, `error`) for live agent-facing
  conversational output.
- [Agent analysis](https://foresea.ink/docs): `POST https://foresea.ink/agent/analyze` runs an
  end-to-end analysis of a live market (fetch price, gather evidence, forecast,
  compute edge) and returns one structured report.
- [Edge scan](https://foresea.ink/docs): `GET https://foresea.ink/agent/scan?platform=polymarket`
  surfaces mispriced live markets (also `kalshi`, or `all`).
- [Radar](https://foresea.ink/radar): daily niche-market radar from Reddit, Polymarket,
  Kalshi, and Foresea's committed live snapshots.
- [PR agent](https://foresea.ink/pr-agent): concise agent-to-agent outreach packet for
  opt-in introductions. It does not send unsolicited messages.
- [OpenAPI spec](https://foresea.ink/openapi.json): full machine-readable API description.

## Track record
- [Live accuracy](https://foresea.ink/track-record): point-in-time forecasts scored
  against prediction-market outcomes, with skill-vs-market by horizon.

## About
- [Web app](https://foresea.ink/): ask any forecasting question in natural language.
- [Source](https://github.com/pareelamre/analyzing-llm-rationale)