Clarify Intent

Prevent intent drift before consequential work. Use available context first, then decide whether to proceed directly, proceed with a lightweight explicit assumption, ask one high-leverage clarification, or request explicit confirmation for high-risk/irreversible actions. Beginner-friendly: plain language, recognition-friendly options, and a "you decide for me" path. Do not interrupt clear, low-risk, reversible tasks just to collect optional preferences. Use whenever a request is vague, has multiple plausible interpretations, may not match the user's real intent, or involves a high-risk or irre

20 stars 1 forksProductivity & Notes

Install

skills CLI (Claude Code, Cursor, Codex and more)
npx skills add Hugo-DDT/clarify-intent --skill clarify-intent
Manual install
git clone --depth 1 https://github.com/Hugo-DDT/clarify-intent.git /tmp/clarify-intent
mkdir -p ~/.claude/skills
cp -r /tmp/clarify-intent/. ~/.claude/skills/clarify-intent

Personal skills live in ~/.claude/skills. Use .claude/skills in a project folder to share a skill with your team.

From the SKILL.md

# Clarify Intent · 意图对齐

## 目标

这个 skill 的目标不是“多问问题”,而是**用最小交互成本避免方向跑偏和返工**。

新手常常无法一次把需求说完整。你的职责不是把用户训练成会写 prompt,而是主动建立“足够开始工作的共同理解”。

**澄清只是手段,不是目的。** 你也可以通过读取现有上下文、采用安全默认值、显式说明假设、先做低成本草稿等方式完成对齐。

## 核心原则

1. **先查,再猜,再问。**
   - 先读当前对话、用户提供的文件、项目约定、已有配置和可用工具结果。
   - 能自己确定的,不要反问用户。

2. **有歧义 ≠ 必须提问。**
   - 只有当不同合理解释会让结果明显不同,或猜错代价较高时,才值得打断用户。

3. **问最高价值问题,不填需求表。**
   - 优先问能一次排除最大错误方向的问题。
   - 如果一个问题无论怎么回答都不会改变下一步,就不要问。

4. **默认一次一个问题。**
   - 根据用户回答动态决定是否需要下一问。
   - 只有当多个问题彼此独立、都很容易回答,而且一次展示明显更省事时,才合并 2–3 个。

5. **可逆的事情大胆推进,不可逆的事情谨慎确认。**
   - 低成本、易修改的任务可以采用合理默认值。
   - 删除、覆盖、付费、对外发送、生产环境修改、数据迁移等高影响动作要显式确认关键事实。

6. **澄清的终点不是“信息完整”,而是“已经足够安全地开始工作”。**

---

## 第 0 步:先利用已有信息

在提问前,先检查:

- 用户是否已经在前文说过答案;
- 文件、代码、表格、配置里是否已经能看出答案;
- 项目是否已有明确技术栈、格式、风格或惯例;
- 是否可以通过工具快速查到;
- 是否存在行业常见且低风险的默认值。

**禁止为了“流程完整”重复询问已知信息。**

---

## 第 1 步:判断是否存在“实质性歧义”

不要因为信息不完整就自动提问。先判断缺失信息是否会改变结果。

一个不确定点值得澄清,通常至少满足下面之一:

- 存在两种或以上都合理的解释,而且会导向明显不同的产出;
- 猜错后返工成本高;
- 任务涉及不可逆、外部、付费、生产环境或数据风险;
- 用户的目标、对象或核心产出物尚不明确,导致无法安全开始;
- 用户的请求建立在疑似错误前提上(如误以为某方案免费、某接口可直接调用),顺着做会走偏;
- 用户明确要求先对齐需求再执行。

内部快速扫描清单(只用于你判断,不要丢给用户):

- 不可度量的修饰词(“快一点”“专业点”“稳一点”);
- 范围边界不清(“所有文件”“整个项目”);
- 隐含假设(“用户”指谁、跑在什么环境);
- 两种都合理的读法;
- 关键信息完全缺失(目标、对象、完成标准都没有);
- 前提可能错误(误以为某功能免费 / 某接口存在 / 某条件成立)。

通常**不值得**为下面这些情况打断用户:

- 只是语气、排版、措辞等容易修改的小偏好;
- 有明显、安全、常见的默认值;
- 可以先产出一个低成本版本再让用户调整;
- 答案可以从现有上下文直接获得。

---

## 第 2 步:在四种行动里选一种

### A. PROCEED — 直接做

适用:意图已经足够清楚,或剩余不确定性不会实质改变结果。

行为:直接执行,不要为了仪式感复述用户的话。

例:
> 用户:把这段中文翻译成英文。
>
> 直接翻译。
Read the full skill on GitHub

Details

Skill name
clarify-intent
License
MIT
Repo updated
18 September 2026

FAQ

What does the Clarify Intent skill do?

Prevent intent drift before consequential work. Use available context first, then decide whether to proceed directly, proceed with a lightweight explicit assumption, ask one high-leverage clarification, or request explicit confirmation for high-risk/irreversible actions. Beginner-friendly: plain language, recognition-friendly options, and a "you decide for me" path. Do not interrupt clear, low-risk, reversible tasks just to collect optional preferences. Use whenever a request is vague, has multiple plausible interpretations, may not match the user's real intent, or involves a high-risk or irre

How do I install the Clarify Intent skill?

skills CLI (Claude Code, Cursor, Codex and more): npx skills add Hugo-DDT/clarify-intent --skill clarify-intent. Skills go in ~/.claude/skills (personal) or .claude/skills (project), and Claude loads them automatically when a task matches the description.

Is Clarify Intent popular?

It lives in Hugo-DDT/clarify-intent, which has 20 GitHub stars and 1 forks.

Related skills

Compaction Advisor

17

vignesh07/compaction-advisor

Provides context-aware compaction guidance with intelligent checkpointing. Monitors context during long tasks and suggests checkpoints before compaction interrupts your work.

Productivity & Notes

Be Ambitious Harness

23

alexpiar/be-ambitious-harness

Harness for ambitious AI task statements. Use when the user writes /be-ambitious, "make this task deeper", "improve the task statement", "raise the bar", "strengthen the prompt" (or in Russian «сделай задачу глубже», «улучши постановку», «поставь амбициознее», «усили промпт») — or before launching any substantial task worth raising the bar on. Routes the task by scenario (design, data, code, writing, decision, generic), runs the analysis, builds a prompt with a gate check, hands it to a worker, runs an independent verifier. Not needed for five-minute tasks.

Productivity & Notes

Filemaker Xml

19

andykear/FileMaker-XMLsnippet-Claude-Skill

Use this skill whenever the user wants to work with FileMaker script XML or custom function XML. This includes: generating FileMaker XML from a description, pseudocode, or script steps; analysing or reviewing existing FileMaker XML for errors or silent-failure risks; converting scripts between formats; or any task where the output must be paste-ready XML for the FileMaker Script Workspace or Manage Custom Functions dialog. Trigger any time the user mentions FileMaker scripts, fmxmlsnippet, Set Variable XML, script step XML, custom function XML, or asks to produce XML that can be pasted into Fi

Productivity & Notes

Travel Search Ru

19

MissiaL/travel-search-ru

Use while planning a trip when Russian-catalog search is needed: package tours, hotels, flights, trains, excursions, prices, and booking links. Trigger on Russian requests: «спланировать путешествие», «подобрать тур», «найти отель/авиабилеты/жд билеты/экскурсии», «маршрут с актуальными ценами». Not for general advice without live search.

Productivity & Notes

Journal Page

19

lumen-prism/journal-page

整页手帐——把一页手帐画成一张图:几张水彩贴纸(淡水彩洇开 + 波浪裁切线)有大有小、歪着放、互相压着,标题压在纸胶带上,宽和纸胶带、票根、横线/方格/牛皮纸卡片、一段草草的英文手写、带虚线箭头的批注、零碎小涂鸦。想"做一页手帐/手帐排版/拼贴一页/画一页手帐"时用。headless Chrome 里 canvas 画,node 出图。

Productivity & Notes

Browser Automation

19

archubbuck/workspace-architect

Local Python-based browser automation toolkit using Playwright. Provides command-line tools for navigating, interacting with, and testing web applications without using MCP protocols. Supports clicking, typing, hovering, screenshots, content extraction, and JavaScript execution.

Productivity & Notes

Browse the most popular Claude skills, all productivity & notes skills, or write your own with the skills generator.