aeondave/verification-before-completion

Verification Before Completion

Completion is a claim. Claims need fresh evidence.

Core rule

Do not say work is complete, fixed, passing, clean, validated, or ready unless the verification command or artifact was checked in the current context.

Gate workflow

1. Name the claim: what exactly is being asserted?
2. Choose proof: command, test, diff, log, replay, artifact, or checklist that would falsify the claim.
3. Run or inspect fresh evidence: use the full relevant check, not a partial proxy.
4. Read the output: exit code, failures, warnings, skipped checks, and scope limits.
5. Report accurately: claim success only when evidence supports it; otherwise state the actual status and next smallest fix.

Common claim gates

| Claim | Requires | Not enough | |---|---|---| | Tests pass | fresh focused or suite output with zero relevant failures | previous run or assumed pass | | Build succeeds | build command exit 0 | lint-only success | | Bug fixed | original reproducer now passes | code changed near symptom | | Requirements met | checklist against spec or request | tests pass without coverage of requirements | | Skill valid | quick_validate.py on changed skill | frontmatter looks right by inspection | | Delegated work done | inspect diff/artifacts and verify outputs | worker report says done |

Security and offensive focus

• Exploit/tooling claims need reproduced primitive, target/build context, or marked uncertainty.
• Cleanup/remediation claims need post-action target-state evidence.
• Scanner triage needs replay, source/sink confirmation, or explicit downgrade.
• Report-ready findings need evidence, limits, and smallest next verification step.

Resources

Load on demand:

• references/completion-evidence.md — detailed claim evidence table and red flags.