Verification Patterns

Purpose

Provide detailed verification procedures for the Agentic Verifier mode. Load this skill when performing thorough verification checks.

When to Use

• •During verification phase (Phase 4)
• •When detailed check procedures are needed
• •When generating verification reports

Detailed Check Procedures

Check 1: Scope Compliance

1. •Read task contract for allowed_files and forbidden_files
2. •Use list_files to enumerate modified files
3. •For each modified file:
   • •Verify it's in allowed_files
   • •Verify it's not in forbidden_files
4. •Flag any unrelated changes (files modified but not in allowed list)

Check 2: Acceptance Criteria

For each criterion in the task contract:

1. •Identify the type: code change, test, documentation, behavior
2. •Find evidence:
   • •Code changes: Specific file:line references
   • •Tests: Test names and pass/fail status
   • •Behavior: Build output, runtime verification
3. •Document in report with PASS/FAIL status
4. •For FAIL: Describe what's missing or incorrect

Check 3: Build Verification

Process:

1. •Use read_file first to check for obvious syntax errors
2. •Execute appropriate build command:
   • •.NET: dotnet build
   • •Node.js: npm run build or tsc
   • •Python: python -m py_compile <files>
3. •Capture output for report
4. •FAIL immediately on build errors (blocking issue)

Common Build Issues:

Check 4: Test Verification

Process:

1. •Identify test commands from:
   • •Plan's Testing Strategy section
   • •Task contract's test requirements
   • •Project conventions (.csproj, package.json)
2. •Run relevant tests (not full suite if scoped):
   • •.NET: dotnet test --filter "FullyQualifiedName~<namespace>"
   • •Node.js: npm test -- --grep "<pattern>"
   • •Python: pytest <path> -k "<pattern>"
3. •Capture results
4. •FAIL on any test failure

Test Coverage Checks:

• •Are new code paths tested?
• •Are edge cases covered?
• •Are error conditions handled?

Check 5: Quality Standards

Check 6: AI Artifact Detection

Use search_files with these patterns:

// TODO.*task
// AI:
// Generated
// task-
// As per spec
// Implementing requirement

DO NOT use grep, findstr, or PowerShell commands for this.

Check 7: Test Completeness

Verification Report Template

# Verification Report: {task-id}

## Result: {PASSED|FAILED|BLOCKED}

## Task Summary
- **Task**: {title}
- **Files Modified**: {count}
- **Quality Bar**: {Standard|High|Critical}

## Acceptance Criteria Status

| # | Criterion | Status | Evidence |
|---|-----------|--------|----------|
| 1 | {criterion text} | PASS/FAIL | {file:line or test name} |
| 2 | {criterion text} | PASS/FAIL | {file:line or test name} |

## Quality Checks

| Check | Status | Notes |
|-------|--------|-------|
| Scope Compliance | PASS/FAIL | {details} |
| Build Verification | PASS/FAIL | {build output summary} |
| Test Verification | PASS/FAIL | {test results summary} |
| Code Quality | PASS/FAIL | {conventions check} |
| AI Artifacts | PASS/FAIL | {found items or "None found"} |

## Issues Found

### Issue #1: {Title}
- **Type**: {missing-test|pattern-violation|etc.}
- **Severity**: blocker|major|minor
- **File**: `path/to/file.cs`
- **Line**: 42-48
- **Description**: {what's wrong}
- **Expected**: {what should be}
- **Suggested Fix**: {how to fix}

## Recommendation

{Next step: PASS → proceed | FAIL → create Q-task | BLOCKED → escalate}

Server Lifecycle for Verification

When verification requires a running server:

Complete Lifecycle

1. •

   START (background):

   • •Windows: start "VerifyServer" cmd /c "npm run dev"
   • •Unix: nohup npm run dev > server.log 2>&1 &
2. •

   WAIT (with timeout):

   cmd

   timeout /t 5 /nobreak >nul
   curl -s http://localhost:3000 >nul 2>&1
   

3. •

   TEST: Run verification checks against running server

4. •

   TERMINATE (guaranteed cleanup):

   • •Windows: taskkill /FI "WINDOWTITLE eq VerifyServer*" /F
   • •Unix: kill $(cat server.pid)
5. •

   RECOVER (if needed):

   • •Windows: taskkill /IM node.exe /F /T
   • •Unix: lsof -ti:3000 | xargs kill -9

Preference Order

1. •Build-only verification (no server) - PREFERRED
2. •Unit tests (no server)
3. •Server verification (only when necessary)
4. •Always clean up servers

Quality Bar Guidance

Standard (Default)

• •Build passes
• •All acceptance criteria met
• •No AI artifacts
• •Conventions followed

High (Public APIs, Breaking Changes)

All of Standard plus:

• •Integration tests pass
• •API documentation complete
• •Backward compatibility verified (or breaking change documented)
• •Code review checklist complete

Critical (Security, Data Integrity)

All of High plus:

• •Security review checklist
• •Data validation verified
• •Rollback plan documented
• •Error handling comprehensive

Anti-Patterns

• •❌ Modifying any source code (verifier is READ-ONLY)
• •❌ Using shell commands for file reading (use built-in tools)
• •❌ Skipping build verification
• •❌ Approving with failing tests
• •❌ Not documenting issues clearly
• •❌ Leaving servers running after verification