将已完成工程化建模的Deep Research报告，编译为可直接用于Obsidian的三层结构化Markdown知识库大纲，严格保留Knowledge ID、学习资源与通关标准的映射关系，适合个人知识管理、Agent系统调用与学习路径搭建。

中文版提示词

请将一份已完成工程化建模的 Deep Research 报告，编译为可直接用于 Obsidian 的结构化 Markdown 知识库大纲。该大纲将作为自动生成笔记、绑定学习资源与通关标准、被 Agent 系统调用的唯一结构蓝图。

输入资料包含全景知识图谱（含层级与 Knowledge ID）、黄金学习资源（每条绑定 Knowledge ID）、通关标准（每条绑定 Knowledge ID）及模块化学习阶段。你必须完整保留并显式映射这些结构关系。

输出规则（强约束）：
1. 只允许出现三种 Markdown 元素：二级标题（##）、三级标题（###）、无序列表项（-）。禁止段落说明、解释性文字、引言、总结、过渡句、表格、代码块。
2. H2 为顶级模块（对应 Layer 1 / Module），须编号，如"## 01-模块名称"。
3. H3 为知识大类（对应 Layer 2），使用名词或名词短语命名。
4. 列表项为原子知识节点，每项严格按以下结构（顺序不可变）：
- [Knowledge ID] 技能名称
  - 学习资源：
    - 资源名称 | 类型 | URL
  - 通关标准：
    - 可验证的产出或能力描述

具体要求：Knowledge ID 须与原报告完全一致；技能名称须为可执行动作；学习资源只用报告已有资源，不得新增、杜撰、泛化；通关标准须对应报告中的 Success Metrics，不得抽象为"理解/掌握/熟悉"。

输出前自检：每个 Knowledge ID 恰好出现一次；每个原子知识点同时绑定资源与通关标准；结构体现"入门→进阶→系统化→工程化"；无资源与知识点脱钩、无通关标准缺技能、无漂浮知识点。
最终只输出 Markdown 大纲，不做任何解释、评价或建议。

英文版提示词

Compile an already engineering-modeled Deep Research report into a structured Markdown knowledge-base outline usable directly in Obsidian. This outline serves as the single structural blueprint for auto-generating notes, binding learning resources and passing criteria, and being invoked by agent systems.

The input is an engineered knowledge asset containing a full knowledge graph (with hierarchy and Knowledge IDs), a golden repository of learning resources (each bound to a Knowledge ID), success metrics (each bound to a Knowledge ID), and modular learning stages. You must fully preserve and explicitly map these structural relationships.

Output rules (strict):
1. Only three Markdown elements are allowed: H2 (##), H3 (###), and unordered list items (-). No paragraphs, explanatory text, intros, summaries, transitions, tables, or code blocks.
2. H2 = top-level module (Layer 1 / Module), numbered, e.g. "## 01-Module Name".
3. H3 = knowledge group (Layer 2), named with nouns or noun phrases.
4. Each list item = an atomic knowledge unit, strictly in this structure (order fixed):
- [Knowledge ID] Skill name
  - Learning resources:
    - Resource name | Type | URL
  - Passing criteria:
    - Verifiable output or capability description

Specifics: Knowledge IDs must match the report exactly; skill names must be executable actions; use only resources already in the report (no invention or generalization); passing criteria must map to the report's Success Metrics and must not be abstracted as "understand/master/familiar".

Self-check before output: each Knowledge ID appears exactly once; each atomic unit binds both resources and passing criteria; structure reflects "beginner → advanced → systematized → engineered"; no detached resources, no criteria without skills, no floating units.
Output only the final Markdown outline, with no explanation, evaluation, or suggestions.

🛠️ **适用 AI 工具**：ChatGPT、Claude、Kimi、DeepSeek、Gemini、通义千问