SDD Audit Skill

Compare implementations against specifications and generate structured review reports.

When to Use

• •Code review before marking tasks complete
• •Quality assurance checks
• •Verifying implementation matches spec
• •Identifying technical debt and issues

Workflow

1. •Load specs: Read spec.md → plan.md → tasks.md
2. •Identify scope: Check todo-list.md for completed work
3. •Inspect code: Review implementation for each task
4. •Gap analysis: Compare spec requirements vs actual code
5. •Generate report: Structured findings with severity levels

Audit Checklist

Reference references/checklist.md for the complete audit checklist covering:

• •Functional requirements compliance
• •Non-functional requirements (performance, security)
• •Code quality standards
• •Edge case handling

Severity Levels

• •CRITICAL: Broken functionality, security risk, release blocker
• •MAJOR: Logic error, missing feature, significant bug
• •MINOR: Style issue, optimization opportunity, cleanup
• •OUTDATED: Code correct but spec needs updating

Report Format

## Audit Report: [Feature]

**Status:** Pass | Fail | Warnings
**Date:** [timestamp]
**Spec Version:** [version]

### Quick Stats
- CRITICAL: [N]
- MAJOR: [N]  
- MINOR: [N]

### Review Comments
| ID | Severity | Location | Issue | Recommendation |
|:--:|:--------:|:---------|:------|:---------------|
| #1 | CRITICAL | `file:line` | Description | Fix suggestion |

### Spec Compliance
| Requirement | Status | Evidence |
|-------------|--------|----------|
| Req 1 | MET | [file:line] |
| Req 2 | NOT MET | [gap description] |

### Recommended Actions
1. [Priority action]
2. [Secondary action]

Integration

• •Works with sdd-verifier subagent for automated validation
• •Can trigger scripts/validate.sh for automated checks
• •Best used with Cursor's Debug Mode for runtime evidence
• •Use the ask question tool if audit criteria are unclear