From personal-harness
Context Budget Monitor — Analyzes context window budget ratios (CLAUDE.md ≤5%, rules ≤3%, MEMORY.md ≤2%, actual work ≥80%) and warns about obesity symptoms. Triggers on 'context budget', 'token ratio'.
How this skill is triggered — by the user, by Claude, or both
Slash command
/personal-harness:context-budgetThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Analyzes context window budget allocation to provide early warning of Context Rot.
Analyzes context window budget allocation to provide early warning of Context Rot.
| Source | Recommended Ratio | Description |
|---|---|---|
| CLAUDE.md | ≤5% | Table-of-contents role, within 100 lines |
| .claude/rules/ | ≤3% | Active rules only |
| MEMORY.md | ≤2% | Within 200 lines |
| Tool definitions | ≤10% | Minimize number of tools |
| Actual work context | ≥80% | Code, tests, error logs |
Estimates approximate token counts for each file (1 token ≈ 4 characters for English, ≈ 1.5 characters for Korean):
Warns when any of the following are detected:
## Context Budget Analysis
### Current Allocation
| Source | Lines | Est. Tokens | Ratio | Status |
|------|------|----------|------|------|
| CLAUDE.md | N | ~T | X% | OK/WARN |
| rules/ | N | ~T | X% | OK/WARN |
| MEMORY.md | N | ~T | X% | OK/WARN |
### Obesity Symptoms
- [WARN] CLAUDE.md 150 lines — reduce to 100 lines or fewer
- [OK] No repeated file-read patterns detected
### Recommended Actions
1. Move detailed content from CLAUDE.md to L1 documents
2. ...
npx claudepluginhub geun-oh/personal-harness --plugin personal-harnessGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.