← ClaudeAtlas

verification-before-completionlisted

Use before claiming ANY task is complete. Requires you to run actual verification commands, read real output, and confirm it matches expectations — before stating the task is done. Blocks "should work" rationalizations.
RBraga01/a-team · ★ 6 · AI & Automation · score 71
Install: claude install-skill RBraga01/a-team
# Verification Before Completion ## The Law ``` YOU CANNOT CLAIM SUCCESS WITHOUT EVIDENCE. "It should work" is not evidence. "I believe it works" is not evidence. Running the command and reading the output IS evidence. ``` ## When to Use Use this skill before saying ANY of these: - "Done" / "Complete" / "Finished" - "The tests pass" - "The build succeeds" - "The feature works" - "The bug is fixed" If you have not run the verification command and read its output in this session, you cannot make the claim. ## The Verification Process ### Step 1: Identify the Verification Command Before claiming completion, state the exact command that proves the claim: | Claim | Verification Command | |-------|---------------------| | "Tests pass" | `npm test` / `pytest` / `go test ./...` / `cargo test` | | "Build succeeds" | `npm run build` / `go build ./...` / `cargo build --release` | | "Type checking passes" | `npx tsc --noEmit` / `mypy .` | | "Bug is fixed" | Run the exact steps that reproduced the bug | | "Feature works" | Run the specific user flow end-to-end | | "Coverage is 80%+" | `npm run test:coverage` / `pytest --cov` | | "No security issues" | `npm audit` / `cargo audit` / `bandit -r .` | ### Step 2: Run the Command Fresh Run it NOW, in this session. Not "I ran it earlier." Not "it was passing before I made changes." **Fresh means:** After the last code change, run the command again. ### Step 3: Read the Actual Output Do not skim. Do not assume. Read: - The exit code