菜单
Use before acting on human instructions: separate what they said from what they meant.
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。
基于 SOC 职业分类
Load for large human-facing outputs: reports, docs, multi-section explanations, artifacts. Choose the right form for each beat (prose, diagram, table, mockup) and put the answer first, depth behind it. If the medium allows, use progressive disclosure.
Use when deciding where knowledge goes or reading/writing durable docs: AGENTS.md, .context/, KB, docs/, and work directories.
Use when exploring or changing the codebase: read AGENTS.md first, use .context/CONTEXT.md for detail, keep intent docs succinct.
Load before writing or revising human-facing text. Choose words deliberately, ground the piece in the reader's context, and remove default LLM phrasing before the final draft.
Use when validating markdown links or Mermaid diagrams.
User-invoked pause before reporting to check intent vs literal completion, surface adjacent wins, and route knowledge capture.
| name | intent-modeling |
| description | Use before acting on human instructions: separate what they said from what they meant. |
Read for the outcome the human wants, not only the words they used. Use conversation context and prior turns to infer underlying goal. When the goal is ambiguous, ask.
Directional corrections usually calibrate a balance ("less X, more Y"), not a permanent ban. Encode the target balance in what you do next.
Match depth and shape to what they need: brief when they want brevity, thorough when they want depth. A vague request may be a rough draft of the real need. Serve the underlying goal, which can be more or less than what feels helpful.
When intent and output diverge, scan other artifacts for the same pattern. One misread often repeats.
