Verification Before Completion
Overview
Claiming work is complete without verification is dishonesty, not efficiency.
Core principle: Evidence before claims, always.
Violating the letter of this rule is violating the spirit of this rule.
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
Common Failures
| 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 |
| PR task complete | GitHub Actions CI green + deployment workflow succeeded | PR merged |
Red Flags โ STOP
- 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
Rationalization Prevention
| Excuse | Reality |
|---|
| "Should work now" | RUN the verification |
| "I'm confident" | Confidence โ evidence |
| "Just this once" | No exceptions |
| "Linter passed" | Linter โ compiler |
| "Agent said success" | Verify independently |
| "I'm tired" | Exhaustion โ excuse |
| "Partial check is enough" | Partial proves nothing |
| "Different words so rule doesn't apply" | Spirit over letter |
Key Patterns
Tests:
โ
[Run test command] [See: 34/34 pass] "All tests pass"
โ "Should pass now" / "Looks correct"
Regression tests (TDD Red-Green):
โ
Write โ Run (pass) โ Revert fix โ Run (MUST FAIL) โ Restore โ Run (pass)
โ "I've written a regression test" (without red-green verification)
Build:
โ
[Run build] [See: exit 0] "Build passes"
โ "Linter passed" (linter doesn't check compilation)
Requirements:
โ
Re-read plan โ Create checklist โ Verify each โ Report gaps or completion
โ "Tests pass, phase complete"
Agent delegation:
โ
Agent reports success โ Check VCS diff โ Verify changes โ Report actual state
โ Trust agent report
Post-merge deployment (GitHub Actions via github MCP):
โ
PR merged โ Check Actions run โ CI green + deploy workflow succeeded โ Task complete
โ "PR merged" (deployment may have failed silently)
Post-Merge Verification
When a task involves a PR that has been merged, "done" means the deployment succeeded โ not just that the code was merged. Use the github MCP (already configured) to check:
1. gh run list --branch main --limit 5 # list recent workflow runs
2. gh run view <run-id> # inspect CI + deploy workflow
3. All jobs: โ
green โ task is complete
Any job: โ failed โ task is NOT complete โ investigate and fix
This costs nothing โ the github MCP is already in .mcp.json. The agent can check deployment status directly without leaving the session.
Trigger: Apply this check whenever:
- A PR was merged as part of the task
- The task description included deployment or release
- The human says "it's merged" or "it's deployed"
Do NOT skip this step by assuming merge = success. CI can pass locally and fail in the pipeline (environment variables, container build, migration runner).
Test Command by Stack
When about to claim tests pass, first confirm you're running the right command. Pick by what files are in the changed set:
| Project marker / changed files | Test command |
|---|
pom.xml or *.java | mvn test -q |
nest-cli.json or *.ts with NestJS imports | npm test |
pyproject.toml or *.py (FastAPI / plain Python) | pytest -q |
pubspec.yaml or *.dart | flutter test |
angular.json or *.ts with Angular imports | ng test --watch=false |
pyproject.toml + LangGraph/LangChain imports | pytest -q |
If multiple markers match (e.g., a monorepo with Java + TypeScript), run the test command for each changed service separately.
Never run the wrong stack's test command โ npm test in a Python project exits 0 with no output, giving false confidence.
Why This Matters
Verification failures lead to:
- Broken trust ("I don't believe you")
- Undefined functions shipped โ would crash
- Missing requirements shipped โ incomplete features
- Time wasted on false completion โ redirect โ rework
This rule exists because honesty is a core value. Unverified claims are not efficiency โ they are debt.
When To Apply
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
Rule applies to:
- Exact phrases
- Paraphrases and synonyms
- Implications of success
- ANY communication suggesting completion/correctness
The Bottom Line
No shortcuts for verification.
Run the command. Read the output. THEN claim the result.
This is non-negotiable.
