audit

Delegate a quality review to the auditor agent.

Steps

Step 1: Determine scope

If the user provided a specific scope (file, task, or area):

• Use that as the audit target

If no scope provided:

• Default to "today's work" — read the daily note for context

Step 2: Select audit tier

Tier	When	Depth
T1	Daily wrap-up, quick check	Scan for obvious issues, 2-3 min
T2	After completing a feature or multi-step task	Check completeness, consistency, side effects
T3	Weekly review, after major changes	Full regression check, knowledge-base sweep
T4	Monthly or after system changes	Deep infrastructure audit, cross-file coherence

Default to T2 for explicit /audit calls.

Step 3: Delegate to auditor

Spawn the auditor agent with the appropriate tier and scope:

Agent(auditor): [Tier] audit of [scope].

Context:
- [Brief description of what was done]
- [Key files involved]

Check for:
1. Completeness — were all requirements met?
2. Consistency — do changes align with existing patterns?
3. Side effects — did changes break anything downstream?
4. Knowledge — are there learnings to promote or nominate?

Report findings as PASS/WARN/FAIL with specific file:line references.

Step 4: Process results

• PASS: Log success, note any suggestions
• WARN: Log warnings, add to daily note
• FAIL: Log failures, add to incident log, create corrective tasks on Task Board

Step 5: Update logs

Append audit results to daily note under a new section:

## Audit — HH:MM (T[tier])
- Result: [PASS/WARN/FAIL]
- Findings: [bullets]
- Actions: [any corrective tasks created]