skill-study-page

これはレビュー時に保存された抜粋です。完全で最新の内容は外部ソースを確認してください。

---
name: skill-study-page
description: Generate a concise, interactive, optionally bilingual HTML learning page that helps non-engineering users understand an Agent Skill from a GitHub URL, raw SKILL.md URL, local SKILL.md file, or pasted Skill content. Use when the user wants to explain, teach, study, translate, compare, or make a visual guide for a Skill, especially content-generation Skills or Skill repositories such as anthropics/skills.
---

# Skill Study Page

## Overview

Turn a Skill into a plain-language study page. Focus on what the Skill does, when to use it, how it works, what the user must provide, what it produces, and where human verification is still needed.

The page can be English-only, Chinese-only, or bilingual. Prefer bilingual output when the user asks for Chinese, translation, language switching, or a page for mixed English/Chinese readers.

Use `references/study-page-schema.md` when preparing the structured content. Use `scripts/render_study_page.py` to render that content into a self-contained HTML page.

## Workflow

1. Resolve the Skill source.
   - For a GitHub repository URL, find candidate `SKILL.md` files or ask which Skill folder to use if the repository contains many.
   - For a GitHub folder URL, convert it to the corresponding raw `SKILL.md` URL when possible.
   - For a raw URL or local path, read the `SKILL.md` directly.
   - For pasted content, use it as the source and label the source as "Pasted Skill content".

2. Read the Skill like a user educator.
   - Extract the frontmatter name and description.
   - Identify the main problem it solves.
   - Identify concrete trigger phrases and situations.
   - Summarize the working strategy as steps, not as implementation trivia.
   - Preserve important tool, file, browser, runtime, or verification dependencies.
   - Note what a non-engineering user must manually check.

3. Create a JSON analysis file.
   - Follow `references/study-page-schema.md`.
   - Keep prose plain, specific, and short.
   - Prefer concrete examples over generic capability lists.
   - Do not invent capabilities that are not implied by the Skill.
   - For bilingual pages, write user-facing fields as `{"en": "...", "zh": "..."}` objects. The renderer will fall back to English if Chinese is missing.

4. Render the page.
   - Run `python3 scripts/render_study_page.py analysis.json output.html`.
   - Open or report the generated HTML path.
   - For bilingual pages, verify the language switcher changes headings, body content, practice tasks, and prompt buttons.
   - If the page is for a specific user workflow, include at least three practice tasks.

## Content Rules

- Write for a capable non-engineer. Avoid unexplained runtime, package, CLI, and protocol jargon.
- Treat the Skill as a work method, not a plugin listing.
- Explain "when to use" and "when not to use" equally.
- Name dependencies and verification needs plainly.
- Include example prompts that a user can copy.
- Translate meaning, not word order. Chinese text should read naturally for Chinese-speaking non-engineering users.
- If the source is ambiguous or a repository contains many Skills, ask the user to choose one before rendering.

## Output Quality Bar

The generated page should feel like a compact product guide, not documentation pasted into a template. It should have a clear headline, language switcher when bilingual content exists, tabs, workflow steps, practice checklist, copyable prompts, and explicit manual verification notes.