# Prompt Refinement

> Refine a prompt by interviewing the user's intent from multiple angles, generating multiple variations at different magnitudes of change, and scoring each against an editorial rubric defined before the refinement starts.

- Skill: `hugobowne/prompt-refinement` (Agent Skill, multi-file: 3 files)
- Install (CLI): `npx skillmds@latest add hugobowne/prompt-refinement`
- Raw SKILL.md: https://api.skillmd.com/api/skills/hugobowne/prompt-refinement/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: hugobowne (https://skillmd.com/u/hugobowne)
- Updated: 2026-09-21
- Page: https://skillmd.com/skills/hugobowne/prompt-refinement

---


> **Note to the agent:** This skill is a stub. Its authoritative `SKILL.md` has not been ported yet. The author, Hilary Mason, externalised a generic version of the skill from Hidden Door's codebase and offered it on the live stream, but flagged that the extraction lost personality and that she plans to edit it before sharing. See this skill's `README.md` for what the skill does, why it is notable, and the timestamped source on the live recording.
>
> If the user wants to use this skill now, do not invent or pretend a procedure exists. Instead, offer to reconstruct a working skill from the `README.md`, adapted to the user's own use case, and tell them plainly that it is your reconstruction, not the author's version.