Since September 2025 this archive has tracked 70 reset-related events, including 7 banked-reset drops, averaging one every 5.9 days. Scope and required action vary by announcement; the table preserves each source so eligibility can be checked rather than assumed.
Download, cite and test the archiveOPEN DATA
JSON includes the schema, full records and event-by-event backtest; CSV is ready for spreadsheets and analysis tools. Both come from the same snapshot.
Backtest method and limits
For event i, the baseline uses only gaps observed before i: previous-event date plus the prior mean gap. The first six events seed the baseline; neither the target nor later events enter its forecast, preventing future-data leakage.
Boundary: This benchmark measures how inaccurate a historical mean can be. It is not a probability model, official schedule or future promise.
Data dictionary
id | Source event identifier. |
|---|---|
announced_at | Announcement timestamp in ISO 8601 UTC. |
date_utc | UTC calendar date derived from announced_at. |
reset_type | Archive label: regular, banked or observed. |
scope_class | Primary scope detected in the announcement; unknown means unstated. |
scope_signals | All scope labels explicitly detected in source text. |
source_author | Author handle recorded by the source archive. |
source_url | Original announcement URL. |
announcement_text | Announcement text preserved by the source archive. |
gap_days_since_previous | Elapsed days since the previous archived event; null for the first row. |
Suggested citation: Codex Reset Radar, “Codex Reset Event Archive,” snapshot 2026-11-01, source codex-resets.com.
All resetsFULL LIST
Browse by era: Launch era · Winter compensation era · March incident cluster · Growth era · Spring tuning era · July reset spree · Sol era · Milestones & banked credits · Astra era