What problem does it solve? AI agents and developers often claim work is done, fixed, or passing without actually running tests or builds, leading to broken commits, false status reports, and lost trust. This Skill imposes a strict evidence-before-claims gate so no completion statement is made without fresh verification output. ## Core Features & Use Cases - Five-Step Gate Function: Identify the proving command, run it fresh, read full output and exit codes, verify the result matches the claim, then state the claim with evidence. - Failure Pattern Catalog: Maps common claims (tests pass, build succeeds, bug fixed, agent completed) to the exact evidence required, plus red-flag phrases like "should work" or "probably fixed" that must stop execution. - Rationalization Prevention: Counters excuses such as "just this once", "linter passed", or "agent said success" with explicit reality checks, including TDD red-green regression verification. - Use Case: Before committing a bug fix or creating a PR, the Skill forces you to run the full test suite, confirm zero failures in the output, and only then report completion with the evidence attached. ## Quick Start Before marking this task done or creating a PR, apply the verification-before-completion gate: run the full test command, read the output, and report the result with evidence.
