菜单
Use when a user request is ambiguous in a way that meaningfully changes the answer. The model asks one focused clarifying question before producing a response, then proceeds.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/nimaaksoy/nimaaksoy.com --skill clarification-first-prompting命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
Use when drafting a cold B2B outreach email. Produces a short message with one specific hook, one line of relevance, and a soft, low-commitment call to action.
Use when the user gives a subject, story, or text and asks for a Persian / Farsi poem. Produces a poem in a chosen classical style (Saadi, Hafez, Rumi, Ferdowsi, Khayyam, modern) with meter and rhyme.
Use when generating prompts for AI video tools (Sora, Runway, Veo, Kling, Pika) to create hypnotic 3–5 second loop videos for music. Produces atmospheric, loop-friendly visual briefs.
正在显示 SKILL.md
| name | Clarification-First Prompting |
| description | Use when a user request is ambiguous in a way that meaningfully changes the answer. The model asks one focused clarifying question before producing a response, then proceeds. |
| dependencies | [] |
| category | meta |
| subcategory | prompt-engineering |
| tags | ["prompting","clarification","ambiguity"] |
| author | {"name":"Nima Aksoy","url":"https://nimaaksoy.com","github":"nimaaksoy"} |
| license | CC-BY-4.0 |
| version | 0.1.0 |
| created | 2026-05-14T00:00:00.000Z |
| updated | 2026-05-14T00:00:00.000Z |
A meta skill that teaches the model to pause and ask one focused clarifying question before answering an ambiguous request. The skill is opinionated about restraint: at most one question, only when the answer would be meaningfully different across plausible readings of the request.
Do not use this skill when the request is clear, when the ambiguity is cosmetic, or when the user has already answered the question implicitly elsewhere in the conversation.
The bar for asking is high. A wrong guess that the user can correct in one turn is often cheaper than a clarifier that breaks their flow.
Input:
User asks: "Write me a short bio."
Expected behaviour: The skill should ask one question before writing. Two interpretations diverge sharply: a professional LinkedIn-style bio versus a personal dating-app-style bio. The model should ask:
Quick check before I draft this: is this for a professional context like a website or LinkedIn, or a personal one like a social profile?
Then write once the user answers.
Input:
User asks: "Write me a 3-sentence professional bio for the about page of my freelance design site. I am Maya, I have ten years of brand work, mostly for food and hospitality clients."
Expected behaviour: The skill should not ask a question. Audience, format, length, and content are all specified. The model writes the bio directly.
Input:
User asks: "Translate this to Spanish."
Expected behaviour: The skill should not ask whether the user wants Latin American or Castilian Spanish. The cost of guessing wrong is one turn; the cost of asking is interrupting the user. Translate using neutral Latin American Spanish and offer a one-line note that it can be adjusted.
None.
0.1.0 — initial version.