my-knowledge-base/ raw/ (your source material - articles, notes, screenshots) wiki/ (where your AI will write the organized version) outputs/ (answers, reports, and research your AI generates)
# Knowledge Base Schema ## What This Is A personal knowledge base about [YOUR TOPIC]. ## How It's Organized - raw/ contains unprocessed source material. Never modify these files. - wiki/ contains the organized wiki. AI maintains this entirely. - outputs/ contains generated reports, answers, and analyses. ## Wiki Rules - Every topic gets its own .md file in wiki/ - Every wiki file starts with a one-paragraph summary - Link related topics to each other using [[topic-name]] format - Maintain an INDEX.md in wiki/ that lists every topic with a one-line description - When new raw sources are added, update the relevant wiki articles ## My Interests [List 3-5 things you want this knowledge base to focus on] 如果你是做咨询的，比如专注在「消费品行业」，可以这样填写 What This Is 和 My Interests： text ## What This Is A knowledge base about consumer goods industry research, brand strategy, and go-to-market best practices. ## My Interests - Brand positioning frameworks - Successful case studies in FMCG - Go-to-market strategies in China and Southeast Asia - Consumer behavior insights for Gen Z
基础设施搭好了，接下来就是系统的灵魂：让 AI 自己把 raw/ 里的杂乱资料，整理成 wiki/ 里的结构化知识。
推荐用支持「读本地文件」的 AI 工具，例如 Claude Code、Cursor、VS Code + AI 插件等。核心思路是：把项目文件夹打开，让 AI 扫描 raw/ 和 CLAUDE.md，然后按照规则生成 wiki/ 内容。
一个典型的提问（Prompt）可以这样写（英文效果通常会更稳定）：
text You are my knowledge base maintainer. 1. Read all files in the raw/ directory. 2. Read the schema and rules in CLAUDE.md. 3. Create or update files in the wiki/ directory according to these rules: - Create an INDEX.md first, listing all major topics with one-line descriptions. - For each major topic, create a separate .md file in wiki/. - Each wiki file must start with a one-paragraph summary. - Use [[topic-name]] links to connect related topics. 4. Do not modify any file in raw/. 5. When you are done, briefly report which wiki files you created or updated.
执行之后，你会在 wiki/ 文件夹里看到一堆新生成的 .md 文件，内容一般会包括：
核心概念的定义和总结
来自不同资料的观点对比
相关主题之间的链接（用 [[xxx]] 的形式）
一个 INDEX.md，帮助你快速浏览整个知识库脉络
从这一步开始，你就拥有了一个由 AI 维护的「活的百科全书」。
第六步：持续提问，并把答案存回系统（长期进行）
有了 Wiki 之后，你要做的就不是「维护知识库」，而是「向自己的知识库提问」。
比如：
text Based on everything in the wiki/ directory, what are the three biggest gaps in my understanding of [某个主题]? 或者： text Compare what source A says about [某个概念] with what source B says. Where do they disagree? Only use information from raw/ and wiki/. 又或者： text Write a 500-word briefing in Chinese on [某个主题], using only the information in this knowledge base (raw/ + wiki/). Structure it with a clear summary and 3 main points.
当 AI 给出答案后，**你可以让它自动把这次输出存进 outputs/**，或者直接更新相关的 wiki/ 文章。
例如，你可以在对话中加一句：
text After generating the answer, save it as a new markdown file in the outputs/ directory, with a descriptive filename.
这样，每一次提问，都会让你的知识库变得更厚、更系统，也更贴合你的工作场景。
第七步：定期做一次「健康体检」（建议每月一次）
因为 AI 也会犯错，如果你完全不检查，错误也会随着知识的积累而不断放大。解决办法很简单：让 AI 自己做周期性的「健康检查」。
可以每月或每季度，给 AI 一个这样的任务：
text Act as a knowledge base auditor. 1. Review all files in the wiki/ directory. 2. Flag any contradictions between articles. 3. Find topics that are mentioned but never properly explained. 4. List any claims that do not have supporting sources in raw/. 5. Suggest 3–5 new wiki articles or sections that would significantly improve coverage and clarity. Return your findings as a markdown report.
粤ICP备14082021号
免费POC，
零成本试错
