# Refine Skill

> Audit and slim a SKILL.md: invocation mode, description, body, references.

- Skill: `mia-cx/refine-skill` (Agent Skill)
- Install (CLI): `npx skillmds@latest add mia-cx/refine-skill`
- Raw SKILL.md: https://api.skillmd.com/api/skills/mia-cx/refine-skill/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Security
- Author: mia-cx (https://skillmd.com/u/mia-cx)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/mia-cx/refine-skill

---


# Refine skill

Apply [agent-copy](../agent-copy/SKILL.md) in full; Read it first if not already loaded. Its pointer rules audit the description, and its skill-mechanics section audits the invocation mode. Checks specific to skills:

- **Check the invocation mode first.** A wrong mode invalidates how the description should read.
- Model-invoked descriptions follow the template `Use when the user asks to …`, in the user's words.
- Examples are concrete and adaptable, not generic.
- Work re-derived across runs (a converter, a query, a validation) becomes `scripts/`.
- Verify tool names, flags, and paths still exist: a skill caches the world and the world moves.

When quality is contested or the description mis-triggers in practice, go empirical: run representative prompts with and without the skill and compare.