菜单
Resolve ambiguities in a feature spec through targeted Q&A. Run after spec-kit-specify when open questions remain. Updates spec.md incrementally.
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/aircury/ai-framework --skill spec-kit-clarify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。
Distil reusable engineering rules from code changes and record them in docs/rules/. Use whenever the user wants to extract rules, conventions, or lessons from a diff, a commit, or the current uncommitted changes — e.g. "extract the rule from these changes", "what convention does this commit establish", "record this as a rule", or after a correction during review that should become a durable guideline for future work.
Read-only cross-artifact consistency check across spec.md, plan.md, and tasks.md. Detects gaps, conflicts, and FRAMEWORK.md violations before implementation begins.
Create a technical implementation plan from a feature spec. Documents architecture, data models, and interface contracts without generating code. Run after spec-kit-clarify.
基于 SOC 职业分类
正在显示 SKILL.md
| name | spec-kit-clarify |
| description | Resolve ambiguities in a feature spec through targeted Q&A. Run after spec-kit-specify when open questions remain. Updates spec.md incrementally. |
| license | MIT |
| compatibility | No external workflow CLI required. |
| metadata | {"author":"Aircury","version":"1.0"} |
Resolve ambiguities in a feature spec through focused, sequential questioning.
Input: Optionally specify a change name. If omitted, infer from context or ask.
Steps
Select the change
Auto-select if only one directory exists under specs/changes/. Otherwise ask.
Load specs/changes/<name>/spec.md
If it does not exist, prompt the user to use the spec-kit-specify skill first.
Scan for ambiguities
Examine the spec for issues across these categories (in priority order):
Generate up to 5 questions total. Skip categories with no ambiguity.
Ask one question at a time
Format each question as multiple-choice (2–5 options) or a short answer (≤5 words).
Only ask a question if the answer materially impacts architecture, data modelling, testing, UX, operations, or compliance.
After each answer, update spec.md before asking the next question.
Finalise
After all questions are resolved (or the 5-question limit is reached), update the spec to:
[NEEDS CLARIFICATION] markers.Output
After completing all questions:
spec-kit-plan skill to create the technical implementation plan."Guardrails
spec-kit-plan skill.