Work Verification

Run verification checklist to confirm work is actually complete. Never trust output alone.

Setup

Before starting: check .handoff/sessions/ for active sessions, read context status.yaml, run git status. Follow .rules/universal.md (Plan -> Approve -> Execute).

<!-- Rules loaded from .rules/universal.md -->

Commands

• /verify task {files...} - Verify task completion
• /verify deploy {team} {service} - Verify deployment
• /verify pr - Verify PR readiness
• /verify tests {path} - Verify tests pass

Core Principle

Agent outputs can describe work without actually performing it. Verification is mandatory.

/verify task {files...}

1. File existence: ls -la {files}
2. File content: Read each file, check syntax, logic, no placeholder TODOs
3. Git status: git status && git diff --stat — verify changes match claims
4. Run tests: php artisan test / npm test / go test / flutter test
5. Report: Files verified table, test results, git status, VERIFIED/NOT VERIFIED

/verify deploy {team} {service}

1. Check deployment platform status (via MCP if configured)
2. Health endpoint: curl -s -o /dev/null -w "%{http_code}" https://{url}/health
3. Check logs for startup errors / exceptions
4. Version verification: curl https://{url}/api/version

/verify pr

1. Branch status: git log origin/main..HEAD --oneline
2. Run tests
3. Run linting
4. Check for debug code: grep -r "console.log\|debugger\|dd(\|dump(" .

Red Flags

• Perfect metrics (100% pass, 95%+ coverage) — suspicious
• Rapid completion of complex work
• Detailed docs without evidence
• No error mentions — real work always has hiccups

Enforcement

No verification = No completion. Status should only show complete after files verified, tests run and passing, git status reviewed.

Full SOP: docs/system/sops/verification-protocol.md

Self-Assessment (Internal)

Rate this skill execution 0-10. If < 8, write feedback to .handoff/feedback/{date}-{skill-name}.yaml with: skill, rating, friction, suggestion, context. Do not mention to user.