# conduct-check - Verify implementation

You are an AI agent using Conduct v0.2 to verify that a spec was implemented correctly.

## Your Task

Verify the implementation thoroughly and document findings in a check log.

## Steps

### 1. Load the Run

```bash
# Find the run
conduct run list

# Get run details
conduct run get {run-id}
```

### 2. Load the Spec

```bash
# Get the spec
conduct spec get {spec-id}
```

Review:
- Acceptance criteria
- Requirements
- Expected outcomes

### 3. Test the Implementation

**Manual Testing:**
- Follow the spec's test plan
- Test happy paths
- Test edge cases
- Test error handling

**Automated Testing:**
```bash
npm test
npm run lint
npm run build
```

**Documentation:**
- Check that code is documented
- Verify README updates
- Check API documentation

### 4. Create Check Log

```markdown
# Check {id}: {Title}

## Meta
- **Check ID:** {id}
- **Run ID:** {run-id}
- **Spec ID:** {spec-id}
- **Result:** pass|fail|partial
- **Agent:** {your-name}
- **Checked:** {ISO-timestamp}

## Test Results

### Automated Tests
- **Unit Tests:** ✓ 45/45 passing
- **Integration Tests:** ✓ 12/12 passing
- **Build:** ✓ Success
- **Linting:** ✓ No errors

### Manual Tests
1. **Login flow:** ✓ Works as expected
2. **Error handling:** ✓ Shows proper messages
3. **Edge cases:** ✓ All handled

### Acceptance Criteria
- [x] Criterion 1
- [x] Criterion 2
- [ ] Criterion 3 (issue found)

## Issues Found

### Issue 1: Password validation missing
- **Severity:** Medium
- **Details:** Min length not enforced
- **Fix:** Add validation in auth controller

## Verdict

**Result:** pass|fail|partial

**Summary:**
{Overall assessment}

**Recommendation:**
- Deploy to staging|Fix issues first|Needs rework
```

### 5. Preview with Dry-Run

```bash
conduct check create _conduct/checks/{id}.v0.check.md \
  --run-id {run-id} \
  --result {pass|fail|partial} \
  --dry-run
```

### 6. Save to Memory

```bash
conduct check create _conduct/checks/{id}.v0.check.md \
  --check-id {id} \
  --run-id {run-id} \
  --result {pass|fail|partial} \
  --agent {your-name}
```

### 7. Follow Up

If issues found:
```bash
# Create a new spec for fixes
conduct-spec

# Or update the spec
conduct spec update {spec-id} --status needs-rework
```

## Output

Present the check summary:

```
✅ Check Complete: CHECK-042

📊 Result: PASS

✓ Tests: All passing
✓ Manual: No issues found
✓ Acceptance: 5/5 met

📝 Recommendation: Ready for deployment

💾 Memory: Check saved to backend
```

## Quick Reference

```bash
# Always dry-run first
conduct check create check.md --run-id RUN-001 --dry-run

# Create check
conduct check create check.md \
  --check-id CHECK-001 \
  --run-id RUN-001 \
  --result pass

# View check
conduct check get CHECK-001

# List checks
conduct check list --run-id RUN-001

# List by result
conduct check list --result fail
```

## Notes

- Always use `--dry-run` first
- Test thoroughly before marking pass
- Document ALL issues found
- Provide actionable recommendations
- Use `partial` for mostly-working implementations
- Use `fail` for broken implementations
- Backend tracks check results for metrics
- Team can see verification status