danfrdn Quality Audit Skill
Use this skill when the user indicates a task is "done" or specifically asks for an audit/review. This is a strict verification step, not a brainstorming session.
Audit Checklist
Assess the code against the following combined standards. If CLI tools (like ruff, bandit, or mypy) are available, use them to validate your findings.
1. Correctness & Functionality
- Correctness: Does the code actually do what the prompt asked?
- Edge cases: Are error conditions (network failures, invalid input, empty states) handled gracefully?
- Logic: Are there any logical fallacies or infinite loops?
2. Security & Integrity
- Security: Are there vulnerabilities (SQLi, XSS, hardcoded secrets)?
- Data Integrity: Are type hints used? Is data validated at boundaries (API inputs/database reads)?
3. Code Health (Cleanliness & Linting)
- Style: Does it follow project conventions (PEP8, etc.)?
- Cleanliness: Is dead code, debug print statements, or commented-out logic removed?
- Linting: Are there unused imports or undefined variables?
4. Simplicity & Maintainability
- Simplicity: Is the code easy to read? (Low Cyclomatic Complexity).
- Best Practices: Is the code modular? Are functions small and single-purpose?
5. Performance
- Efficiency: Are there obvious inefficiencies (e.g., N+1 queries, nested loops on large data)?
- Resource Usage: Are connections/files closed properly?
How to Report
- Status: Start with a strictly formatted status:
AUDIT RESULT: [PASS / FAIL / WARN]. - The "Fix-it" List: For every failure, provide:
- The file/line number.
- The specific violation (e.g., "Security: Hardcoded API key found").
- A corrected code snippet (refactored version).
- Tone: Be objective and critical. Do not compliment the code; focus on hardening it.