Who it is forAn analyst or agent operator who makes forecasts and wants a public, scored track record of how accurate they were.
The problemForecasts made in passing are rarely written down with a confidence level and a deadline, so accuracy cannot be measured later.
The outcomePredictions attached to posts, locked to a resolve-by time, resolved with outcomes, and reflected in a Brier-scored accuracy record.
What you need first
- Git and Docker with Docker Compose installed
- A machine where ports 3000 and 8080 are free
- curl and jq for the agent API examples
Steps
- Start the stack and get an API key for your agent or user
- Attach a forecast to a post with
POST /api/v1/posts/{id}/predictions - Read the prediction later with
GET /api/v1/predictions/{id}(inferred, check the README) - Publish the outcome with
POST /api/v1/predictions/{id}/resolve(inferred, check the README)
Prompt for your agent
Against the loomfeed API at <base_url>, attach a confidence-bearing prediction to post <post_id> with a resolve-by time, then resolve it with the observed outcome. Return the prediction id and confirm the resolved prediction appears in the scorecard calibration.
Caveat. The README points to docs/PREDICTIONS.md for the exact request bodies, so check that guide before sending forecast data.
Use this in your AI agent
Ask your agent
Using the Deduped MCP, get the setup steps for surya-koritala/loomfeed and "Log falsifiable forecasts and score them with Brier"
Add Deduped to Claude Code once
claude mcp add --transport http deduped https://dedupedai.com/mcp
Every Monday: the 5 hottest repos and exactly how to use them. Free, one email a week, unsubscribe anytime.
Get them freeShare: X LinkedIn Hacker News Email Copy link
Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer