A public evidence tool for Codex users

Know when a reset signal is actually useful.

ResetPulse watches public Codex usage-reset signals, records when they were observed, and keeps the public event separate from the private state of your own account.

01

Know when the signal appeared

Server-side checks run about once per minute, even when your computer is off. Source time and first-observed time remain separate in UTC.

02

See why the result is trusted

Each event has a readable evidence page, a reproducible JSON integrity receipt with a public schema, and an explicit distinction between early signal, confirmation, context, and unverified observation.

03

See rollout reality

An event-linked anonymous layer records account or plan observations against the same public signal. It makes rollout mismatches visible without turning an anecdote into global proof.

04

Keep personal quota private

Paste a sanitized read-only usage snapshot into the browser-only parser. It shows your own windows locally, rejects token-like fields, and never uploads account data.

05

Compare before and after

Paste two sanitized captures locally to see whether a balance, used percentage, or reset time changed after a public signal. The comparison never leaves your browser.

Try it without signup

The live timeline, forecast, evidence pages, RSS feed, scope/type classifications, status endpoints, developer badge, and personal browser-only tools are available immediately. Email alerts are optional, and ResetPulse never asks for a Codex password, session, or private usage credential.

Try the live monitor · Check a private snapshot · Read the signal-versus-outcome method · Embed the public evidence · Verify and copy a public receipt

What makes it different

A public announcement, a monitor observation, and an account-level change can be different facts. ResetPulse makes that disagreement inspectable instead of hiding it behind a single confident-looking yes or no. Forecast pages also expose sample size, backtest accuracy, and Brier score rather than presenting an unexplained probability.