git:20260830.0196be9 to git:20260905.f7824ba
34 added, 99 removed. Audit A to A.
---
name: verification-before-completion
- description: Use when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions always
+ description: "Assess nontrivial completion evidence."
---
# Verification Before Completion
- ## Overview
-
- **Core principle:** Evidence before claims, always.
-
- **Violating the letter of this rule is violating the spirit of this rule.**
-
- ## The Iron Law
-
- ```
- NO COMPLETION CLAIMS WITHOUT FRESH VERIFICATION EVIDENCE
- ```
-
- If you haven't run the verification command in this message, you cannot claim it passes.
-
- ## The Gate Function
-
- ```
- BEFORE claiming any status or expressing satisfaction:
-
- 1. IDENTIFY: What command proves this claim?
- 2. RUN: Execute the FULL command (fresh, complete)
- 3. READ: Full output, check exit code, count failures
- 4. VERIFY: Does output confirm the claim?
- - If NO: State actual status with evidence
- - If YES: State claim WITH evidence
- 5. ONLY THEN: Make the claim
-
- Skip any step = lying, not verifying
- ```
+ ## Core principle
- ## Common Failures
+ Evidence before claims. Verification should be fresh, decisive, and proportionate to the result being asserted. For an obvious low-impact local edit, apply the direct check without loading this skill or creating a formal verification artifact. Use this fuller workflow when the evidence choice, failure interpretation or completion boundary is nontrivial, or when explicitly invoked.
- | Claim | Requires | Not Sufficient |
- |-------|----------|----------------|
- | Tests pass | Test command output: 0 failures | Previous run, "should pass" |
- | Linter clean | Linter output: 0 errors | Partial check, extrapolation |
- | Build succeeds | Build command: exit 0 | Linter passing, logs look good |
- | Bug fixed | Test original symptom: passes | Code changed, assumed fixed |
- | Regression test works | Red-green cycle verified | Test passes once |
- | Agent completed | VCS diff shows changes | Agent reports "success" |
- | Requirements met | Line-by-line checklist | Tests passing |
+ ## Lightweight gate
- ## Red Flags - STOP
+ Before claiming that work is complete, fixed, passing, deployed, sent, published, or otherwise in a verified state:
- - Using "should", "probably", "seems to"
- - Expressing satisfaction before verification ("Great!", "Perfect!", "Done!", etc.)
- - About to commit/push/PR without verification
- - Trusting agent success reports
- - Relying on partial verification
- - Thinking "just this once"
- - Tired and wanting work over
- - **ANY wording implying success without having run verification**
+ 1. Name the exact claim and its consequence.
+ 2. Choose the smallest fresh check that can actually prove it.
+ 3. Run the check and read the exit status, failures, and material output.
+ 4. Report the result with exact coverage and anything not verified.
- ## Rationalization Prevention
+ Do not replace a required runtime check with source inspection, a build with lint, or end-to-end evidence with an agent's success message. Do not run an entire test universe when a targeted regression plus the repository's required gate proves the claim.
- | Excuse | Reality |
- |--------|---------|
- | "Should work now" | RUN the verification |
- | "I'm confident" | Confidence ≠ evidence |
- | "Just this once" | No exceptions |
- | "Linter passed" | Linter ≠ compiler |
- | "Agent said success" | Verify independently |
- | "I'm tired" | Exhaustion ≠ excuse |
- | "Partial check is enough" | Partial proves nothing |
- | "Different words so rule doesn't apply" | Spirit over letter |
+ ## Match evidence to the claim
- ## Key Patterns
+ | Claim | Decisive evidence |
+ |---|---|
+ | Targeted bug fixed | Reproduce the original symptom or run its regression test |
+ | Tests pass | The named test command exits successfully with zero relevant failures |
+ | Build succeeds | The actual build command exits successfully |
+ | Visual issue fixed | Inspect the rendered affected state at relevant viewports |
+ | Requirements met | Check each material acceptance criterion against code, output, or runtime evidence |
+ | External state changed | Read back the exact account, message, deployment, or service state |
+ | Research complete | State sources searched, semantic coverage, exclusions, and unresolved limits |
- **Tests:**
- ```
- ✅ [Run test command] [See: 34/34 pass] "All tests pass"
- ❌ "Should pass now" / "Looks correct"
- ```
+ ## Coverage language
- **Regression tests (TDD Red-Green):**
- ```
- ✅ Write → Run (pass) → Revert fix → Run (MUST FAIL) → Restore → Run (pass)
- ❌ "I've written a regression test" (without red-green verification)
- ```
+ Use precise labels such as:
- **Build:**
- ```
- ✅ [Run build] [See: exit 0] "Build passes"
- ❌ "Linter passed" (linter doesn't check compilation)
- ```
+ - source-inspected;
+ - structurally verified;
+ - targeted test passed;
+ - runtime-tested;
+ - visually inspected;
+ - external state confirmed;
+ - untested or blocked.
- **Requirements:**
- ```
- ✅ Re-read plan → Create checklist → Verify each → Report gaps or completion
- ❌ "Tests pass, phase complete"
- ```
+ Never imply whole-system certification from a sample. Previous evidence may provide context, but a current completion claim needs current verification when the state could have changed.
- **Agent delegation:**
- ```
- ✅ Agent reports success → Check VCS diff → Verify changes → Report actual state
- ❌ Trust agent report
- ```
+ ## Delegated work
- ## When To Apply
+ A subagent report is evidence about what the subagent observed, not automatic proof of completion. Inspect the relevant diff or artifact and run the smallest decisive verification before adopting its claim.
- **ALWAYS before:**
- - ANY variation of success/completion claims
- - ANY expression of satisfaction
- - ANY positive statement about work state
- - Committing, PR creation, task completion
- - Moving to next task
- - Delegating to agents
+ ## Stop condition
- **Rule applies to:**
- - Exact phrases
- - Paraphrases and synonyms
- - Implications of success
- - ANY communication suggesting completion/correctness
+ Stop when the acceptance criteria are supported at the required confidence. Record residual limitations instead of adding ceremonial checks that cannot change the decision.