Falsifying

Attacks a green signal to find out whether it can go red at all. Audits the measuring apparatus — a gate, checker, coverage number, CI job, or test suite — rather than the product: constructs the case that must fail and checks that it does, and diffs every independent source of the same fact against the others. Use when a gate or checker is written or changed, when a metric looks better than the work feels, before trusting a number in an MR, or when a suite passes on code nobody has exercised. Produces a failing case or a source-disagreement table as evidence; a confirmed finding is handed to the BUG flow, not fixed in place. Not for a reported bug with a symptom (that is `diagnosing-bugs`), not a diff review (`code-review`), not a hunt through product code for latent defects (`bug-hunter`), and not live HTTP abuse probes (`attack-test`).

witooh Updated 0 repo stars

File contents

witooh/neo-plugin commit fd31214bb0

Frequently asked questions

npx get-skill add witooh/falsifying