workspace-audit

name	workspace-audit
description	Self-audit workspace configuration for consistency — verify modes, rules, skills alignment and detect issues.
risk	safe
source	self
tags	["audit","configuration","consistency"]

Workspace Audit

When to Use

• When user requests "audit workspace" or "check configuration"
• After adding/modifying modes, rules, or skills
• Periodically (recommend monthly) to catch drift

Instructions

Step 1: Modes Audit

• List all modes from .roomodes and settings customModes
• Verify each mode has: roleDefinition, whenToUse, groups
• Check groups restrict file access appropriately per mode purpose
• Flag: modes without whenToUse, modes with overly permissive groups

Step 2: Rules Audit

• List all rules from .roo/rules/ (global) and .roo/rules-{mode}/ (mode-specific)
• Check for conflicts (rule A says "always X", rule B says "never X")
• Check total rule content size — warn if >2000 lines (context bloat risk)
• Flag: duplicate rules, empty rule files, rules that contradict each other

Step 3: Skills Audit

• List all skills from .roo/skills/
• Verify each has YAML frontmatter with name, description
• Cross-check with skills-registry.md — all skills listed?
• Cross-check with skill-awareness.md — all skills in awareness table?
• Flag: skills not in registry, registry entries without matching skill files

Step 4: Settings Audit

• Check modeApiConfigs — all modes have model assignment?
• Check deniedCommands — reasonable security posture?
• Check maxWorkspaceFiles, maxOpenTabsContext — reasonable for project size?
• Flag: permissive settings without justification

Step 5: Generate Report

Output a markdown report:

## Workspace Audit Report — {date}
### Summary: {PASS|WARN|FAIL} ({N} issues)
### Modes: {count} modes, {issues}
### Rules: {count} rules, {issues}
### Skills: {count} skills, {issues}
### Settings: {issues}
### Recommendations: {actionable items}

Important

• This is a READ-ONLY audit — do NOT modify any files
• Report findings objectively — let user decide what to fix
• Prioritize findings: CRITICAL > WARN > INFO