Skill Creator 技能建立指南

來源狀態

已棄用

官方 README 建議改看 Plugins catalog；不表示此 Skill 已一對一搬遷。

資源類型

Skill

與類似 Skill 有什麼不同？

• Agent Skill 編寫與漸進揭露

  Skill Creator 提供技能建立與迭代流程；Agent Skill Authoring 是格式與組織方式的官方文件指南。

來源描述的能力

這些是來源描述的可能操作，不代表 2lus 已授予權限或已測試。

尚未列出能力，請審查原始來源。

這個 Skill 是什麼？

協助設計 Skill 的任務邊界、說明與附屬資源，避免把整份知識庫塞進每次都載入的指引。

可以做什麼？何時適合使用？

• 重複執行專案交付流程。
• 把團隊規則變成可套用的任務指引。
• 改善既有 Skill 的觸發條件與資源結構。

如何使用

1. 整理真實輸入、預期輸出與不該執行的動作。
2. 將核心指引放入 SKILL.md，大型資料分至 references，重複操作再考慮 scripts。
3. 用代表性案例人工檢查觸發與安全邊界，依目前官方文件封裝。

使用前你需要準備

• 具體工作範例
• 目標 Agent 與既有技能結構

你可以替換：

[workflow]、[agent]

環境與相依需求

• 明確的任務案例與支援 Skills 的 Codex 環境。

使用範例與 Prompt

以下是 2lus 撰寫的示範需求；請替換為你有權處理的檔案與專案，不代表已執行或保證結果。

入門

使用 Skill Creator，把每週發布檢查整理成 Skill；先問我必要輸入、輸出與禁止動作。

實務

為資料清理流程設計 SKILL.md、references 與 scripts 的分工，不執行腳本，列出人工驗收案例。

進階

審查現有 Skill 是否過度觸發、引用失效或授權範圍過大；提出最小修改與回歸檢查清單。

實用提醒

• 先做一個範圍小、可驗收的流程，不要建立萬用助手。

限制與注意事項

• 舊 openai/skills repository 已標示 deprecated；封裝與安裝請依新版官方文件，不假定所有版本內建此技能。

安全注意事項

• 先審查來源與相依工具，只處理獲准的資料；密鑰、個資與登入狀態不得放入提示或公開輸出。

使用與設定

此條目不提供已確認的通用安裝指令；請依官方文件與 Agent 版本操作。

支援平台

Codex — 文件記載，未做實機相容性測試 官方文件 ↗

來源與授權

來源查核日期（非安全認證）: 2026-09-29

請參考原始來源授權條款

相關 Skills

• Find Skills 技能探索
• Playwright 網站測試

Source status

Deprecated

The official README recommends the Plugins catalog; this does not establish a one-to-one relocation of this skill.

Resource type

Skill

How is this different from similar skills?

• Agent Skill Authoring & Progressive Disclosure

  Skill Creator provides a creation and iteration workflow; Agent Skill Authoring is official documentation about format and organization.

Documented capabilities

These are operations described upstream, not permissions granted or tested by 2lus.

Capabilities not declared here; review the original source.

What is this skill?

Define task boundaries and resource structure without loading an entire knowledge base into every invocation.

Use cases and when to use it

• Repeat a release workflow.
• Apply team conventions consistently.
• Improve triggers and resources in an existing skill.

How to use it

1. List real inputs, outputs and prohibited actions.
2. Keep essential instructions in SKILL.md; separate references and reusable scripts.
3. Review representative cases and safety boundaries, then use current packaging documentation.

What you need

• Concrete workflow examples
• Target agent and existing skill structure

You can replace:

[workflow], [agent]

Environment and dependencies

• Concrete workflow examples and a Codex environment supporting skills.

Usage and prompt examples

These example requests were written by 2lus. Substitute files and projects you may use; examples are not executed results or guarantees.

Beginner

Use Skill Creator to capture my weekly release review. Ask about required inputs, outputs and prohibited actions first.

Practical

Design SKILL.md, references and scripts for a data-cleanup workflow. Do not execute scripts; propose manual acceptance cases.

Advanced

Review this skill for over-broad triggers, broken references and excessive permissions. Propose minimal edits and regression checks.

Tips

• Start with one bounded, reviewable workflow.

Limitations

• The old openai/skills repository is deprecated. Consult current packaging guidance; do not assume every version bundles this skill.

Security notes

• Review sources/dependencies and process only authorized data; keep credentials, personal data and session state out of prompts and public outputs.

Usage and setup

No verified universal installation command is provided for this entry. Follow the official documentation for your agent version.

Supported agents

Codex — Documented; not runtime-tested Documentation ↗

Sources and license

Source check date (not a safety certification): 2026-09-29

Refer to the original source license terms

Related skills

• Find Skills
• Playwright