quality-verify-integration

📁 dawiddutoit/custom-claude 📅 Jan 26, 2026
4
总安装量
4
周安装量
#48393
全站排名
安装命令
npx skills add https://github.com/dawiddutoit/custom-claude --skill quality-verify-integration

Agent 安装分布

mcpjam 4
neovate 4
gemini-cli 4
antigravity 4
windsurf 4
zencoder 4

Skill 文档

Verify Integration

Purpose

Prevent “done but not integrated” failures where:

  • Code exists ✓
  • Unit tests pass ✓
  • Quality gates pass ✓
  • But code is never called at runtime ✗

When to Use

MANDATORY before:

  • Moving ADR to 3_completed/
  • Marking todo tasks complete for integration work
  • Claiming a feature is “done” or “working”
  • Creating a PR for integration changes

The CCV Principle

COMPLETE = CREATION + CONNECTION + VERIFICATION
Phase What It Proves Required Evidence
CREATION Artifact exists File, tests, types
CONNECTION Wired into system Import, registration
VERIFICATION Works at runtime Logs, output

Missing any phase = NOT complete

The Four Questions Test

Before “done”, answer ALL FOUR:

  1. How do I trigger this? (entry point)
  2. What connects it to the system? (import/registration)
  3. What proves it runs? (logs/traces)
  4. What shows it works? (outcome)

Cannot answer all four? → NOT COMPLETE

Quick Verification

Step 1: Check for Orphaned Modules

# Run orphan detection script
./scripts/verify_integration.sh

# Or manual check for specific module
grep -r "from.*module_name import\|import.*module_name" src/ --include="*.py" | grep -v test

If no matches → Module NOT integrated

Step 2: Verify Call-Sites

# Check if function/class is actually called
grep -r "function_name\|ClassName" src/ --include="*.py" | grep -v "^def \|^class " | grep -v test

If no matches → Code never called

Step 3: Check Context-Specific Integration

LangGraph Nodes:

grep -n "from.*nodes import\|add_node.*name" src/temet_run/coordination/graph/builder.py

DI Services:

grep -n "providers.Singleton\|providers.Factory" src/temet_run/container.py

CLI Commands:

grep -n "add_command\|@app.command" src/temet_run/cli/app.py

Step 4: Demand Runtime Proof

Require one of:

  • Integration test output showing component exists
  • Logs from execution showing code ran
  • State inspection showing fields populated

Connection Patterns

See references/connection-patterns.md for:

  • LangGraph node integration
  • Dependency Injection patterns
  • CLI command registration
  • API endpoint wiring
  • Configuration loading

Verification Report Template

## Integration Verification: [Feature Name]

### CCV Status

**CREATION:** ✅ / ❌
- Files: [list]
- Tests: [count] passing
- Types: mypy passes

**CONNECTION:** ✅ / ❌
- Import location: [file:line]
- Registration: [how wired]
- Entry point: [how triggered]

**VERIFICATION:** ✅ / ❌
- Integration test: [pass/fail]
- Runtime logs: [attached/missing]
- Expected outcome: [observed/not observed]

### Four Questions

1. Trigger: [answer or UNANSWERED]
2. Connection: [answer or UNANSWERED]
3. Execution proof: [answer or UNANSWERED]
4. Outcome proof: [answer or UNANSWERED]

### Verdict

**APPROVED ✅** — All phases complete, evidence attached
OR
**BLOCKED ❌** — Missing: [list what's missing]

Supporting Files

Success Criteria

  • No orphaned modules (verify_integration.sh passes)
  • All imports verified with grep
  • All call-sites verified
  • Context-specific checks passed
  • Four Questions answered
  • Runtime proof attached