# Unhobble

> Bare-baseline experiment: strip a repo's standing instructions, log stumbles against the bare model, then restore only those with repeated same-cause evidence. Measures the model where audit-instructions judges the text. Use when: 'unhobble', 'run the bare experiment', 'delete my CLAUDE.md and see', 'does the model still need these instructions', 'new model dropped, re-baseline', 'instruction ablation experiment', 'deletion watch', 'watch this rule before deleting it'. Human-gated, resumable.

- Skill: `melodic-software/unhobble-3` (Agent Skill)
- Install (CLI): `npx get-skill add melodic-software/unhobble-3`
- Raw SKILL.md: https://api.skillmd.com/api/skills/melodic-software/unhobble-3/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: melodic-software (https://skillmd.com/u/melodic-software)
- Updated: 2026-11-12
- Page: https://skillmd.com/skills/melodic-software/unhobble-3

---


# Unhobble (eval stand-in)

Human-gated: every phase waits for the person's go-ahead before it changes any file. This stand-in
runs no phase; say what the experiment would do and stop.

