Datasets:
| { | |
| "name": "track-record", | |
| "version": "1.0.0", | |
| "description": "A calibration loop for AI agents - log every checkable judgment as a bet with a numeric probability and an observable trigger, force settlement against real outcomes, distill recurring biases into rules. Learns from reality vs. its own predictions, not user corrections. Use when making a checkable prediction or judgment call, logging a bet before acting, settling or backfilling past judgments against real outcomes, running a retrospective or calibration review, or whenever track record, calibration, prediction log, settle bets, or learning from outcomes comes up. 中文触发：下判断要登记、复盘、回填、结算、判断校准、校准回顾、赌局台账。", | |
| "author": { "name": "LucioLiu", "url": "https://github.com/LucioLiu" }, | |
| "homepage": "https://github.com/LucioLiu/track-record", | |
| "repository": "https://github.com/LucioLiu/track-record", | |
| "keywords": ["claude-code", "agent-skills", "calibration", "prediction-ledger", "forced-settlement", "self-improving"], | |
| "license": "PolyForm-Noncommercial-1.0.0" | |
| } | |