AU
audit
Run a comprehensive audit of a specific area of the codebase using 3 parallel expert agents. Use when asked to audit, review deeply, or check quality of a module/page/feature.
Install
mkdir -p .claude/skills/audit-alo2118 && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/14669" && unzip -o skill.zip -d .claude/skills/audit-alo2118 && rm skill.zipInstalls to .claude/skills/audit-alo2118
Activation
This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.
Run a comprehensive audit of a specific area of the codebase using 3 parallel expert agents. Use when asked to audit, review deeply, or check quality of a module/page/feature.175 chars✓ has a “when” trigger
About this skill
Run a comprehensive audit of: $ARGUMENTS
Process
1. Dispatch 3 parallel agents (mandatory)
Launch these 3 agents simultaneously using the Agent tool:
- bug-hunter agent — Analyze the target area for bugs, data loss risks, Supabase query safety, race conditions
- ux-reviewer agent — Check UX consistency, accessibility, style constant usage, sibling comparison
- arch-reviewer agent — Check architecture, file ownership, performance, convention compliance
Each agent must receive:
- The specific area/files to analyze
- Context about what was recently changed (if applicable)
2. Consolidate findings
Merge results from all 3 agents into a single ranked list:
- CRITICAL — Data loss, crashes, security issues
- MAJOR — Incorrect behavior, accessibility violations, broken patterns
- MINOR — Style inconsistencies, minor improvements
3. Fix all issues, starting from critical
4. Re-audit after fixes (mandatory)
Run the 3 agents again on the fixed code to verify nothing was missed and no regressions introduced.
5. Build check
Run npm run build — must pass with zero errors.