메뉴
Read-only orientation for a new agent session in a learning project. Use when the host prompts "请使用 project-onboarding skill 了解这个项目". Read project state, summarize in markdown, do NOT modify any files. Run once per session (typically at init).
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Generate a structured reading guide for an academic paper markdown file. Use when the host invokes "请用 typora-paper-reader skill 为以下论文生成导读" with a paper file path and user persona. Reads the paper, identifies key points based on the persona, writes a JSON guide to disk.
Generate review cards (quiz questions + key points) for each concept in a completed learning chapter. Use when the host prompts with chapter content and concept list. Outputs structured JSON only — does not write files.
Conduct a Socratic dialogue to deepen understanding of a concept cluster. Use when the host prompts "请使用 typora-socratic-review skill 进行苏格拉底复习". Returns one Socratic question per turn in markdown; end session with `[SESSION_END]` marker when the user demonstrates mastery or asks to stop.
Explain selected text from a learning chapter using analogies and plain language, then suggest follow-up questions for deeper understanding. Use when the host invokes "请用 explanation skill 解释" with a term, context, and optional previous Q&A history. Writes the result to disk using the Write tool.
Generate quiz questions from saved Cornell explanation cues (concept + Q&A history). Use when the host invokes "请用 extra-quiz-generation skill 根据以下概念列表生成附加题" with a JSON list of concepts and their Q&A histories. Writes the result to disk using the Write tool.
Generate a complete learning chapter (Markdown + quiz.json + concepts.json) for one slot in a learning project outline. Use when the host prompts "请使用 chapter-generation skill 生成第 N 章" with chapter index/title/duration/concepts. Writes three files via the Write tool. Skill is read-only and self-contained — does not call other skills.
| name | project-onboarding |
| description | Read-only orientation for a new agent session in a learning project. Use when the host prompts "请使用 project-onboarding skill 了解这个项目". Read project state, summarize in markdown, do NOT modify any files. Run once per session (typically at init). |
Orient yourself in a learning project workspace. Read-only. Do not modify files.
Host prompt will include the project path. If $ARGUMENTS.project_path is given, use it; otherwise the agent SDK's cwd is the project root.
Use Read and Glob tools only. No Write, no Edit, no Bash.
{project_path}/.learning/project.json. Expect fields: name, created, chapters[], total_duration. If missing, the project hasn't been initialized — tell the user and stop.{project_path}/*.md to find chapter files (zero-padded, e.g. 00-*.md, 01-*.md). Sort by name to get generation order.*.concepts.json to extract top concepts. If none, the project is fresh.Return a single markdown text response (the host displays it as the agent's first message):
## 项目摘要
**主题**: {project.name}
**进度**: 已完成 {N}/{M} 章 ({percent}%)
**当前焦点**: {latest chapter title or "尚未开始"}
**已学概念**: {top 3-5 from concepts.json, comma-separated; "无" if fresh}
**接下来**: {one-line suggestion: 完成第 N+1 章 / 复习 / 等等}
Keep it under 200 words. No code blocks, no Mermaid.
project.json missing → respond 项目尚未初始化 and exit*.md files yet → say 已规划 {M} 章但尚未生成任何内容