Validate Build

Run all automated verification:

Phase 2: Functional Validation (Browser)

For each user flow in Product-Plan:

1. Navigate to entry point
2. Execute flow steps
3. Verify expected outcomes
4. Screenshot key states
5. Report pass/fail with evidence

Phase 3: Visual Regression (Optional)

Compare rendered screens against Preview-Output mockups:

• Flag significant differences
• Require Owner review for discrepancies

Phase 4: Owner Sign-off

Present validation report. Owner must explicitly approve.

Automated Check Execution

# 1. Full test suite
bundle exec rspec

# 2. Coverage check (from SimpleCov output)
cat coverage/.last_run.json
# Check: covered_percent >= 90

# 3. Security scan
bundle exec brakeman --no-pager

# 4. Dependency audit
bundle exec bundle-audit check --update

# 5. Linting
bundle exec rubocop

Functional Flow Template

For each flow in Product-Plan:

## Flow: [Flow Name]

**Product-Plan Reference:** §[section]

### Steps

1. Navigate to [entry point]
   - **Expected:** [what should appear]
   - **Actual:** [what appeared]
   - **Status:** PASS | FAIL

2. [Action]
   - **Expected:** [result]
   - **Actual:** [result]
   - **Status:** PASS | FAIL

### Evidence
- Screenshot: `validation/screenshots/flow-name-step-1.png`

### Verdict: PASS | FAIL

Coverage Gap Handling

If coverage < 90%:

1. Identify gaps (SimpleCov report)
2. Flag high-risk gaps
3. Create test tasks (NOT auto-generate weak tests)
4. Document in validation report

Why not auto-generate? Coverage < 90% indicates TDD wasn't followed. Auto-generating tests hides the process failure.

Auto-Fix Loop (Phase 1)

For fixable issues:

1. Run check
2. If FAIL:
   a. Analyze error
   b. Generate fix
   c. Apply fix
   d. Re-run check
   e. If FAIL after 3 attempts → escalate
3. If PASS → proceed

Escalation Criteria

Escalate to Owner when:

• Auto-fix loop exhausted (3 attempts)
• Security finding requires architectural change
• Functional flow cannot complete
• Coverage gap in complex/risky code
• Visual regression on critical screens

Validation Report Format

Output: docs/validation/YYYY-MM-DD-<product>-report.md

# Validation Report: [Product Name]

**Date:** YYYY-MM-DD
**Plan:** YYYY-MM-DD-HH-MM_product-name
**Status:** PASSED | FAILED | PASSED_WITH_WARNINGS

## Summary

| Category | Result |
|----------|--------|
| Tests | 142 passed, 0 failed |
| Coverage | 94.2% |
| Security | 0 issues |
| Dependencies | 0 vulnerabilities |
| Linting | 0 offenses |
| Functional Flows | 8/8 passed |

## Automated Checks

### Test Suite

142 examples, 0 failures Finished in 12.34 seconds

### Coverage
| Area | Coverage |
|------|----------|
| Models | 98% |
| Controllers | 92% |
| Services | 95% |
| **Total** | **94.2%** |

#### Coverage Gaps
| File | Coverage | Risk |
|------|----------|------|
| (none above threshold) | | |

### Security Scan

== Brakeman Report == No warnings found

### Dependency Audit

No vulnerabilities found

### Linting

142 files inspected, no offenses detected

## Functional Flows

### Flow 1: User Checkout
**Status:** PASS
**Steps:** 7/7 passed
**Evidence:** `validation/screenshots/checkout-*.png`

### Flow 2: Order Management
**Status:** PASS
**Steps:** 5/5 passed
**Evidence:** `validation/screenshots/orders-*.png`

## Visual Regression
[Optional: differences from mockups]

## Recommendations
- [ ] Consider adding integration tests for payment edge cases
- [ ] Monitor performance of order list with large datasets

## Owner Sign-off Required

- [ ] I have reviewed this validation report
- [ ] I accept these results
- [ ] I approve for deployment

**Signed:** _________________ **Date:** _____________

Failure Handling

Test Failure

1. Document failing tests
2. Attempt auto-fix (max 3)
3. If fixed: re-run full suite
4. If not fixed: escalate with diagnosis

Security Finding

1. Classify severity (critical/high/medium/low)
2. If critical/high: block validation
3. Create fix task
4. Re-run after fix

Functional Failure

1. Screenshot failure state
2. Document expected vs actual
3. Create bug fix task
4. Re-validate after fix

Authority

Validate-Build is ADVISORY with BLOCKING POWER:

• Cannot auto-deploy or auto-merge
• CAN prevent plan from being marked complete
• Owner MUST sign off on validation report

Exit Criteria

Before declaring product ready:

• All Automated Checks Pass: Tests, security, linting
• Coverage ≥ 90%: Or gaps documented and accepted
• All Functional Flows Pass: Evidence provided
• No Critical Security Issues: Or documented accepted risks
• Validation Report Complete: All sections filled
• Owner Sign-off: Explicit approval on report

Sub-Domain Reference

• automated-checks.md — Detailed check procedures
• functional-testing.md — Browser testing patterns
• coverage-analysis.md — Coverage gap handling

Integration

Previous skill: Execute-Build (provides implementation)

Backtrack to: Execute-Build (if fixes needed)

Output: Validated, Owner-approved product ready for deployment