AI usage limit reset times and history
This is a rule-based check. No model is called. Don't paste passwords, tokens, private keys or unredacted customer data.
Five things: the goal (what you should get when it's done), scope (which files or directories may be changed, and which must not), risky permissions (whether deleting, overwriting, publishing, committing or installing has an approval step), stop conditions (what to do on a conflict, sensitive data or growing scope) and verification (how to prove it's done). For each one, the checker only looks for matching wording in the prompt.
If you write only the first prompt below, the checker shows 3 of 5 clear, with scope and stop condition missing. Add those two, as in the second prompt, and all five are clear.
Check this project and fix the problems
Fix the failing login only in the src/ directory and run the tests when you are done. Don't change dependencies; if a config change is needed, pause and ask me first.
It only matches common phrasing. It doesn't call a model and can't tell whether a prompt is any good. A “Clear” on one item just means the prompt mentions it. For anything risky, read the prompt yourself once more.
The check runs in your browser and the prompt you paste is never uploaded. Still, don't paste passwords, tokens, private keys or unredacted customer data.