What problem does it solve?
Users of ChatGPT and Codex often cannot tell whether their usage quota was actually reset, when the next reset lands, or whether an unannounced platform-wide reset occurred. This Skill distinguishes four reset types (official Tibo announcements, silent platform resets, banked credit resets, and per-account weekly cycles) and produces a single evidence-backed conclusion instead of ambiguous guesses.
Core Features & Use Cases
- Announcement tracking: Queries the Tibo Radar timeline API, fxtwitter mirrors, and OpenAI status pages to find official reset announcements and incident-driven compensations, with fallback channels when endpoints fail.
- Local forensic analysis: Parses ~/.codex rollout session snapshots to reconstruct weekly quota curves, pinpoint reset intervals to minute-level ranges, and detect silent resets or window reconfigurations.
- Evidence-graded conclusions: Classifies findings by scope (single account, cross-account observation, large-scale silent reset, or officially confirmed global reset) and converts Pacific Time announcements to Beijing time.
- Use Case: A user says their Codex weekly quota suddenly returned to 100% but no tracker shows a reset. The Skill checks local rollout snapshots, searches concurrent user reports, and concludes whether a silent platform reset occurred.
Quick Start
Ask the assistant when my Codex usage limit resets and whether a silent reset already happened today.