---
description: Claude Code settings optimization and token audit
---

# Claude Code Settings Optimizer

Self-audit and optimize Claude Code configuration for maximum token efficiency and performance.

## Triggers
- "settings audit", "exceed response limit", "MCP error"
- "token usage", "context budget"

## Quick Commands

```bash
python3 scripts/run_audit.py
python3 scripts/analyze_tokens.py
python3 scripts/auto_fix.py --apply
```

## Audit Scope

### Token Budget Targets

| Component | Target | Max |
|-----------|--------|-----|
| CLAUDE.md | 2,000 | 3,000 |
| MCP Servers | 5,000 | 10,000 |
| Skills metadata | 500 | 1,000 |
| Working space | >140,000 | -- |

### MCP Server Strategy

| Strategy | Criteria |
|----------|----------|
| `always` | Essential for daily work (Playwright, Notion) |
| `lazy` | Occasionally needed (GitHub, Slack) |
| `disable` | Rarely used or token-heavy |

Every MCP server MUST have `serverInstructions` for Tool Search.

### CLAUDE.md Health Checks
- Line count (warn >200), token estimate (warn >3,000)
- Structure quality, redundant info, unnecessary repetition

### Auto-Fix (safe, with backup)
- Add `serverInstructions` to MCP servers
- Add frontmatter to commands missing it
- Suggest CLAUDE.md compression

## Output Report
- Token budget breakdown and percentages
- Health status (Good/Needs Attention/Critical)
- Findings (Critical/Warnings/Passing)
- Prioritized recommendations