Audit Command

Args: $ARGUMENTS (optional: target path, or "list", "targets", "stats")

Run

node scripts/audit-tracker.cjs $ARGUMENTS

Workflow

Read .claude/rules/auditing.md for the complete 8-dimension audit protocol.

Default Behavior (no args)

1. •Run node scripts/audit-tracker.cjs to get the queue
2. •Immediately propose the top recommendation with a brief rationale
3. •Ask ONE question: "Audit [target]?" with options: Yes / Pick different target
4. •On confirmation, claim and begin the audit

Minimize interaction. User runs /audit, you propose a target, they say go, you audit.

With Target Specified

If user provides a target (e.g., /audit features/compose), skip proposal and begin immediately after claiming.

Large Module Protection

Modules with >30 files or >3 sub-features cannot be audited as a single unit. The script will refuse and show sub-features to audit instead.

Example: features/create (521 files, 6 sub-features) -> audit features/create/assemble, features/create/generate, etc. instead.

If the claim command shows "MODULE TOO LARGE FOR SINGLE AUDIT":

1. •Pick one of the listed sub-features
2. •Claim and audit that sub-feature
3. •Repeat for other sub-features as needed

This ensures audits are thorough and meaningful - you can't claim to have audited 500+ files in one pass.

The Audit Process

1. •Claim the target: node scripts/audit-tracker.cjs claim "<target>"
2. •Read the entry component and map the component tree
3. •Grade across 8 dimensions (see auditing.md):
   • •Architecture, Code Quality, Svelte 5 Compliance, Accessibility
   • •UX States, UI Consistency, Performance, Security
4. •Present the scorecard with grades and prioritized issues
5. •Immediately record grades (no confirmation needed): node scripts/audit-tracker.cjs record "<target>" --grades "A+,A,A,B,A,A,A+,A" --notes "..."
6. •If fixes needed: Present the fix plan and ask "Ready to bring this to A+?" with options to proceed or defer
7. •If no fixes needed (A+ across the board): Ask "Shall I audit the next feature?"

Useful Commands

# See what needs auditing
node scripts/audit-tracker.cjs

# List all recorded audits
node scripts/audit-tracker.cjs list

# List all auditable targets
node scripts/audit-tracker.cjs targets

# Check status of a specific target
node scripts/audit-tracker.cjs status features/compose

# Show coverage stats
node scripts/audit-tracker.cjs stats

# Claim a target (prevents parallel agent conflicts)
node scripts/audit-tracker.cjs claim features/compose

# Force claim a large module (not recommended - prefer sub-features)
node scripts/audit-tracker.cjs claim features/create --force

# Record audit results (order: Arch, Code, Svelte5, A11y, UX, UI, Perf, Security)
node scripts/audit-tracker.cjs record features/compose --grades "A+,A,A,B,A,A,A+,A"

# Release a claim without recording (if audit abandoned)
node scripts/audit-tracker.cjs release features/compose

Post-Audit Workflow

After recording an audit:

1. •Recording auto-releases the claim - no manual release needed
2. •If you made code changes, offer to commit them:
   • •Use /commit or ask user if they want changes committed
   • •Group audit fixes into a single commit with descriptive message
3. •Summary: Provide a brief summary of grades and key changes made

Important Notes

• •Always claim before auditing to prevent conflicts with parallel agents
• •Claims use file locking to prevent race conditions between parallel agents
• •Large modules (>30 files or >3 sub-features) must be audited as sub-features - the script will refuse to claim them
• •Grade order for --grades flag: Architecture, Code Quality, Svelte 5, Accessibility, UX States, UI Consistency, Performance, Security
• •Targets are auto-browseed from src/lib/features/ and src/lib/shared/
• •Sub-features are auto-browseed within large modules (e.g., create/assemble, create/generate)
• •Audits older than 30 days are flagged as stale
• •Claims expire after 4 hours if not released
• •The goal is A+ across all dimensions