sdd-audit

📁 madebyaris/spec-kit-command-cursor 📅 12 days ago
1
总安装量
1
周安装量
#47185
全站排名
安装命令
npx skills add https://github.com/madebyaris/spec-kit-command-cursor --skill sdd-audit

Agent 安装分布

antigravity 1

Skill 文档

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  • OUTDATED: Code correct but spec needs updating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