菜单
Use when creating new skills, editing existing skills, or verifying skills before deployment in Codex CLI.
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
You MUST use this before any creative work - creating features, building components, adding functionality, or modifying behavior. Explores user intent, requirements and design before implementation.
Use when the user asks to review a pull request or diff with a structured Codex CLI workflow using parallel review passes and filtered actionable findings.
Use when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies.
Use when you have a written implementation plan to execute in a separate session with review checkpoints
Use when the user asks to build a new feature and wants a structured Codex CLI workflow with discovery, clarifying questions, architecture comparison, implementation, and review.
Use when implementation is complete, all tests pass, and you need to decide how to integrate the work - guides completion of development work by presenting structured options for merge, PR, or cleanup
| name | writing-skills |
| description | Use when creating new skills, editing existing skills, or verifying skills before deployment in Codex CLI. |
Writing skills is TDD for process documentation: establish a failing baseline, write the minimum skill content that fixes the behavior, then refine it until the workflow is reliable.
Personal skills live in ~/.codex/skills.
If you did not observe a realistic failure mode before writing the skill, you do not yet know what the skill needs to prevent.
A skill is a reusable guide for patterns, tools, or workflows that future Codex sessions should be able to discover and apply.
Skills are:
Skills are not:
ext skills/ skill-name/ SKILL.md supporting-file.md
Use a flat namespace and keep supporting files only when the main file would otherwise become too heavy.
ame uses letters, numbers, and hyphens only
Future Codex sessions need to find the skill quickly. Optimize for that by:
Mirror this checklist in update_plan when actively authoring a skill:
| Mistake | Fix |
|---|---|
| Description explains what the skill does | Rewrite it to explain when to load it |
| Frontmatter includes extra keys | Keep only |
| ame and description | |
| Long narrative examples | Replace with compact reusable examples |
| Supporting docs loaded by default | Move heavy references into separate files |
| No baseline failure observed | Reproduce a realistic failing scenario first |
Use subagents or realistic prompts to test:
The goal is not just "looks good," but "changes behavior predictably."
