学习这个LLM wiki: [https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f](https://gist.github.com/karpath y/442a6bf555914893e9891c11519de94f) 帮我在我的仓库里搭建一套一样的知识库,输出全部用中文。
● Let me first fetch the gist content and explore your vault structure.
● Fetch(https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f)
Listing 1 directory… (ctrl+o to expand) ⎿ (PATH’ 2>/dev/null):{CLAUDE_CONFIG_DIR:-$HOME/.claude}”;
Listed 1 directory (ctrl+o to expand) ⎿ PostToolUse:Bash hook returned blocking error ⎿ [export PATH=“SHELL -lc ‘echo PATH”; _C=“{CLAUDE_CONFIG_DIR:-HOME/.claude}”; _E=S={CLAUDE_PLUGIN_ROOT:-{PLUGIN_ROOT:-}}"; _P=({ [ -n “_E" ] && printf '%s\n' "_E”; ls -dt “_C/plugins/marketplaces/thedotmack/plugin”; } | while IFS= read -r _R; do _R=“{_R%/}"; [ -d "_R/plugin/scripts” ] && _Q=“_R”; [ -f “_Q/scripts/bun-runner.js" ] && [ -f "_Q/scripts/worker-service.cjs” ] && { printf ‘%s\n’ “_Q"; break; }; done); [ -n "_P” ] || { echo “claude-mem: plugin scripts not found” >&2; exit 1; }; command -v cygpath >/dev/null 2>&1 && { _W=_P” 2>/dev/null); [ -n “_W" ] && _P="_W”; }; node
- Nebulizing… (10s · ↑ 255 tokens) “_P/scr pts/worker-service.cjs” hook claude-code ─────────────────────────────────────────────────────────────────observation]: /usr/bin/bash: line 1: printf: write ──── ❯ error: Permission denied ─ ─ ─────────────────────────────── ● WebFetchn被网络限制了,让我换个方式获取 gist 内容。 ● high · /effort
● Web Search(“karpathy LLM wiki gist github “llm-wiki” knowledge base 442a6bf555914893e9891c11519de94f”) ⎿ Did 1 search in 30s ⎿ PostToolUse:WebSearch hook returned blocking error ⎿ [export PATH=“SHELL -lc ‘echo PATH”; _C=“{CLAUDE_CONFIG_DIR:-HOME/.claude}”; _E=“{CLAUDE_PLUGIN_ROOT:-{PLUGIN_ROOT:-}}”; _P=({ [ -n "_E” ] && printf ‘%s\n’ “_C/plugins/cache/thedotmack/claude-mem”/[0-9]*/ 2>/dev/null; printf ‘%s\n’ ”_C/plugins/marketplaces/thedotmack/plugin"; } | while IFS= read -r _R; do _R="{_R%/}”; [ -d “_R/plugin/scripts" ] && _Q="_R/plugin” || _Q=“_Q/scripts/bun-runner.js” ] && [ -f “_Q/scripts/worker-service.cjs" ] && { printf '%s\n' "_Q”; break; }; done); [ -n “_P" ] || { echo "claude-mem: plugin scripts not found" >&2; exit 1; }; command -v cygpath >/dev/null 2>&1 && { _W=(cygpath -w “_W” ] && _P=“_W"; }; node "_P/scripts/bun-runner.js” ”$_P/scripts/worker-service.cjs” hook claude-code observation]: /usr/bin/bash: line 1: printf: write error: Permission denied
● Bash(curl -sL “https://api.github.com/gists/442a6bf555914893e9891c11519de94f” 2>/dev/null | head -500) ⎿ Interrupted · What should Claude do instead?
❯ # LLM Wiki
A pattern for building personal knowledge bases using LLMs.
This is an idea file, it is designed to be copy pasted to your own LLM Agent (e.g. OpenAI Codex, Claude Code, OpenCode / Pi, or etc.). Its goal is to communicate the high level idea, but your agent will build out the specifics in collaboration with you.
The core idea
Most people’s experience with LLMs and documents looks like RAG: you upload a collection of files, the LLM retrieves relevant chunks at query time, and generates an answer. This works, but the LLM is rediscovering knowledge from scratch on every question. There’s no accumulation. Ask a subtle question that requires synthesizing five documents, and the LLM has to find and piece together the relevant fragments every time. Nothing is built up. NotebookLM, ChatGPT file uploads, and most RAG systems work this way.
The idea here is different. Instead of just retrieving from raw documents at query time, the LLM incrementally builds and maintains a persistent wiki — a structured, interlinked collection of markdown files that sits between you and the raw sources. When you add a new source, the LLM doesn’t just index it for later retrieval. It reads it, extracts the key information, and integrates it into the existing wiki — updating entity pages, revising topic summaries, noting where new data contradicts old claims, strengthening or challenging the evolving synthesis. The knowledge is compiled once and then kept current, not re-derived on every query.
This is the key difference: the wiki is a persistent, compounding artifact. The cross-references are already there. The contradictions have already been flagged. The synthesis already reflects everything you’ve read. The wiki keeps getting richer with every source you add and every question you ask.
You never (or rarely) write the wiki yourself — the LLM writes and maintains all of it. You’re in charge of sourcing, exploration, and asking the right questions. The LLM does all the grunt work — the summarizing, cross-referencing, filing, and bookkeeping that makes a knowledge base actually useful over time. In practice, I have the LLM agent open on one side and Obsidian open on the other. The LLM makes edits based on our conversation, and I browse the results in real time — following links, checking the graph view, reading the updated pages. Obsidian is the IDE; the LLM is the programmer; the wiki is the codebase.
This can apply to a lot of different contexts. A few ──── (42 lines hidden) ────────────────────────────────────────────────────────────────────────────────────────────── o what, which pages are hubs, which are orphans.
- Marp is a markdown-based slide deck format. Obsidian has a plugin for it. Useful for generating presentations directly from wiki content.
- Dataview is an Obsidian plugin that runs queries over page frontmatter. If your LLM adds YAML frontmatter to wiki pages (tags, dates, source counts), Dataview can generate dynamic tables and lists.
- The wiki is just a git repo of markdown files. You get version history, branching, and collaboration for free.
Why this works
The tedious part of maintaining a knowledge base is not the reading or the thinking — it’s the bookkeeping. Updating cross-references, keeping summaries current, noting when new data contradicts old claims, maintaining consistency across dozens of pages. Humans abandon wikis because the maintenance burden grows faster than the value. LLMs don’t get bored, don’t forget to update a cross-reference, and can touch 15 files in one pass. The wiki stays maintained because the cost of maintenance is near zero.
- Personal: tracking your own goals, health, psychology, self-improvement — filing journal entries, articles, podcast notes, and building up a structured picture of yourself over time.
- Research: going deep on a topic over weeks or months — reading papers, articles, reports, and incrementally building a comprehensive wiki with an evolving thesis.
- Reading a book: filing each chapter as you go, building out pages for characters, themes, plot threads, and how they connect. By the end you have a rich companion wiki. Think of fan wikis like Tolkien Gateway — thousands of interlinked pages covering characters, places, events, languages, built by a community of volunteers over years. You could build something like that personally as you read, with the LLM doing all the cross-referencing and maintenance.
- Business/team: an internal wiki maintained by LLMs, fed by Slack threads, meeting transcripts, project documents, customer calls. Possibly with humans in the loop reviewing updates. The wiki stays current because the LLM does the maintenance that no one on the team wants to do.
- Competitive analysis, due diligence, trip planning, course notes, hobby deep-dives — anything where you’re accumulating knowledge over time and want it organized rather than scattered.
Architecture
There are three layers:
Raw sources — your curated collection of source documents. Articles, papers, images, data files. These are immutable — the LLM reads from them but never modifies them. This is your source of truth.
The wiki — a directory of LLM-generated markdown files. Summaries, entity pages, concept pages, comparisons, an overview, a synthesis. The LLM owns this layer entirely. It creates pages, updates them when new sources arrive, maintains cross-references, and keeps everything consistent. You read it; the LLM writes it.
The schema — a document (e.g. CLAUDE.md for Claude Code or AGENTS.md for Codex) that tells the LLM how the wiki is structured, what the conventions are, and what workflows to follow when ingesting sources, answering questions, or maintaining the wiki. This is the key configuration file — it’s what makes the LLM a disciplined wiki maintainer rather than a generic chatbot. You and the LLM co-evolve this over time as you figure out what works for your domain.
Operations
Ingest. You drop a new source into the raw collection and tell the LLM to process it. An example flow: the LLM reads the source, discusses key takeaways with you, writes a summary page in the wiki, updates the index, updates relevant entity and concept pages across the wiki, and appends an entry to the log. A single source might touch 10-15 wiki pages. Personally I prefer to ingest sources one at a time and stay involved — I read the summaries, check the updates, and guide the LLM on what to emphasize. But you could also batch-ingest many sources at once with less supervision. It’s up to you to develop the workflow that fits your style and document it in the schema for future sessions.
Query. You ask questions against the wiki. The LLM searches for relevant pages, reads them, and synthesizes an answer with citations. Answers can take different forms depending on the question — a markdown page, a comparison table, a slide deck (Marp), a chart (matplotlib), a canvas. The important insight: good answers can be filed back into the wiki as new pages. A comparison you asked for, an analysis, a connection you discovered — these are valuable and shouldn’t disappear into chat history. This way your explorations compound in the knowledge base just like ingested sources do.
Lint. Periodically, ask the LLM to health-check the wiki. Look for: contradictions between pages, stale claims that newer sources have superseded, orphan pages with no inbound links, important concepts mentioned but lacking their own page, missing cross-references, data gaps that could be filled with a web search. The LLM is good at suggesting new questions to investigate and new sources to look for. This keeps the wiki healthy as it grows.
Indexing and logging
Two special files help the LLM (and you) navigate the wiki as it grows. They serve different purposes:
index.md is content-oriented. It’s a catalog of everything in the wiki — each page listed with a link, a one-line summary, and optionally metadata like date or source count. Organized by category (entities, concepts, sources, etc.). The LLM updates it on every ingest. When answering a query, the LLM reads the index first to find relevant pages, then drills into them. This works surprisingly well at moderate scale (~100 sources, ~hundreds of pages) and avoids the need for embedding-based RAG infrastructure.
log.md is chronological. It’s an append-only record of what happened and when — ingests, queries, lint
passes. A useful tip: if each entry starts with a consistent prefix (e.g. ## [2026-04-02] ingest | Article Title), the log becomes parseable with simple unix tools — grep "^## \[" log.md | tail -5 gives you the
last 5 entries. The log gives you a timeline of the wiki’s evolution and helps the LLM understand what’s been
done recently.
Optional: CLI tools
At some point you may want to build small tools that help the LLM operate on the wiki more efficiently. A search engine over the wiki pages is the most obvious one — at small scale the index file is enough, but as the wiki grows you want proper search. qmd is a good option: it’s a local search engine for markdown files with hybrid BM25/vector search and LLM re-ranking, all on-device. It has both a CLI (so the LLM can shell out to it) and an MCP server (so the LLM can use it as a native tool). You could also build something simpler yourself — the LLM can help you vibe-code a naive search script as the need arises.
Tips and tricks
- Obsidian Web Clipper is a browser extension that converts web articles to markdown. Very useful for quickly getting sources into your raw collection.
- Download images locally. In Obsidian Settings → Files and links, set “Attachment folder path” to a fixed
directory (e.g.
raw/assets/). Then in Settings → Hotkeys, search for “Download” to find “Download attachments for current file” and bind it to a hotkey (e.g. Ctrl+Shift+D). After clipping an article, hit the hotkey and all images get downloaded to local disk. This is optional but useful — it lets the LLM view and reference images directly instead of relying on URLs that may break. Note that LLMs can’t natively read markdown with inline images in one pass — the workaround is to have the LLM read the text first, then view some or all of the referenced images separately to gain additional context. It’s a bit clunky but works well enough. - Obsidian’s graph view is the best way to see the shape of your wiki — what’s connected to what, which pages are hubs, which are orphans.
- Marp is a markdown-based slide deck format. Obsidian has a plugin for it. Useful for generating presentations directly from wiki content.
- Dataview is an Obsidian plugin that runs queries over page frontmatter. If your LLM adds YAML frontmatter to wiki pages (tags, dates, source counts), Dataview can generate dynamic tables and lists.
- The wiki is just a git repo of markdown files. You get version history, branching, and collaboration for free.
Why this works
The tedious part of maintaining a knowledge base is not the reading or the thinking — it’s the bookkeeping. Updating cross-references, keeping summaries current, noting when new data contradicts old claims, maintaining consistency across dozens of pages. Humans abandon wikis because the maintenance burden grows faster than the value. LLMs don’t get bored, don’t forget to update a cross-reference, and can touch 15 files in one pass. The wiki stays maintained because the cost of maintenance is near zero.
The human’s job is to curate sources, direct the analysis, ask good questions, and think about what it all means. The LLM’s job is everything else.
The idea is related in spirit to Vannevar Bush’s Memex (1945) — a personal, curated knowledge store with associative trails between documents. Bush’s vision was closer to this than to what the web became: private, actively curated, with the connections between documents as valuable as the documents themselves. The part he couldn’t solve was who does the maintenance. The LLM handles that.
Note
This document is intentionally abstract. It describes the idea, not a specific implementation. The exact directory structure, the schema conventions, the page formats, the tooling — all of that will depend on your domain, your preferences, and your LLM of choice. Everything mentioned above is optional and modular — pick what’s useful, ignore what isn’t. For example: your sources might be text-only, so you don’t need image handling at all. Your wiki might be small enough that the index file is all you need, no search engine required. You might not care about slide decks and just want markdown pages. You might want a completely different set of output formats. The right way to use this is to share it with your LLM agent and work together to instantiate a version that fits your needs. The document’s only job is to communicate the pattern. Your LLM can figure out the rest.请根据以上内容,在我的仓库里搭建一套相同结构的中文知识库。
● UserPromptSubmit operation blocked by hook: [export PATH=“SHELL -lc ‘echo PATH”; _C=“{CLAUDE_CONFIG_DIR:-HOME/.claude}”; _E=“{CLAUDE_PLUGIN_ROOT:-{PLUGIN_ROOT:-}}”; _P=({ [ -n "_E” ] && printf ‘%s\n’ “_C/plugins/cache/thedotmack/claude-mem”/[0-9]*/ 2>/dev/null; printf ‘%s\n’ ”_C/plugins/marketplaces/thedotmack/plugin"; } | while IFS= read -r _R; do _R="{_R%/}”; [ -d “_R/plugin/scripts" ] && _Q="_R/plugin” || _Q=“_Q/scripts/bun-runner.js” ] && [ -f “_Q/scripts/worker-service.cjs" ] && { printf '%s\n' "_Q”; break; }; done); [ -n “_P" ] || { echo "claude-mem: plugin scripts not found" >&2; exit 1; }; command -v cygpath >/dev/null 2>&1 && { _W=(cygpath -w “_W” ] && _P=“_W"; }; node "_P/scripts/bun-runner.js” ”$_P/scripts/worker-service.cjs” hook claude-code session-init]: /usr/bin/bash: line 1: printf: write error: Permission denied
Original prompt: 开始做了吗
❯ /plugin disable claude-mem ⎿ ✓ Disabled claude-mem. Run /reload-plugins to apply.
❯ /reload-plugins ⎿ Reloaded: 3 plugins · 0 skills · 5 agents · 1 hook · 0 plugin MCP servers · 0 plugin LSP servers
❯ # LLM Wiki
A pattern for building personal knowledge bases using LLMs.
This is an idea file, it is designed to be copy pasted to your own LLM Agent (e.g. OpenAI Codex, Claude Code, OpenCode / Pi, or etc.). Its goal is to communicate the high level idea, but your agent will build out the specifics in collaboration with you.
The core idea
Most people’s experience with LLMs and documents looks like RAG: you upload a collection of files, the LLM retrieves relevant chunks at query time, and generates an answer. This works, but the LLM is rediscovering knowledge from scratch on every question. There’s no accumulation. Ask a subtle question that requires synthesizing five documents, and the LLM has to find and piece together the relevant fragments every time. Nothing is built up. NotebookLM, ChatGPT file uploads, and most RAG systems work this way.
The idea here is different. Instead of just retrieving from raw documents at query time, the LLM incrementally builds and maintains a persistent wiki — a structured, interlinked collection of markdown files that sits between you and the raw sources. When you add a new source, the LLM doesn’t just index it for later retrieval. It reads it, extracts the key information, and integrates it into the existing wiki — updating entity pages, revising topic summaries, noting where new data contradicts old claims, strengthening or challenging the evolving synthesis. The knowledge is compiled once and then kept current, not re-derived on every query.
This is the key difference: the wiki is a persistent, compounding artifact. The cross-references are already there. The contradictions have already been flagged. The synthesis already reflects everything you’ve read. The wiki keeps getting richer with every source you add and every question you ask.
You never (or rarely) write the wiki yourself — the LLM writes and maintains all of it. You’re in charge of sourcing, exploration, and asking the right questions. The LLM does all the grunt work — the summarizing, cross-referencing, filing, and bookkeeping that makes a knowledge base actually useful over time. In practice, I have the LLM agent open on one side and Obsidian open on the other. The LLM makes edits based on our conversation, and I browse the results in real time — following links, checking the graph view, reading the updated pages. Obsidian is the IDE; the LLM is the programmer; the wiki is the codebase.
This can apply to a lot of different contexts. A few ──── (42 lines hidden) ────────────────────────────────────────────────────────────────────────────────────────────── e hubs, which are orphans.
- Marp is a markdown-based slide deck format. Obsidian has a plugin for it. Useful for generating presentations directly from wiki content.
- Dataview is an Obsidian plugin that runs queries over page frontmatter. If your LLM adds YAML frontmatter to wiki pages (tags, dates, source counts), Dataview can generate dynamic tables and lists.
- The wiki is just a git repo of markdown files. You get version history, branching, and collaboration for free.
Why this works
The tedious part of maintaining a knowledge base is not the reading or the thinking — it’s the bookkeeping. Updating cross-references, keeping summaries current, noting when new data contradicts old claims, maintaining consistency across dozens of pages. Humans abandon wikis because the maintenance burden grows faster than the value. LLMs don’t get bored, don’t forget to update a cross-reference, and can touch 15 files in one pass. The wiki stays maintained because the cost of maintenance is near zero.
The human’s job is to curate sources, direct the analysis, ask good questions, and think about what it all means. The LLM’s job is everything else.
The idea is related in spirit to Vannevar Bush’s Memex (1945) — a personal, curated knowledge store with associative trails between documents. Bush’s vision was closer to this than to what the web became: private, actively curated, with the connections between documents as valuable as the documents themselves. The part he couldn’t solve was who does the maintenance. The LLM handles that.
Note
This document is intentionally abstract. It describes the idea, not a specific implementation. The exact directory structure, the schema conventions, the page formats, the tooling — all of that will depend on your domain, your preferences, and your LLM of choice. Everything mentioned above is optional and modular — pick what’s useful, ignore what isn’t. For example: your sources might be text-only, so you don’t need image handling at all. Your wiki might be small enough that the index file is all you need, no search engine required. You might not care about slide decks and just want markdown pages. You might want a completely different set of output formats. The right way to use this is to share it with your LLM agent and work together to instantiate a version that fits your needs. The document’s only job is to communicate the pattern. Your LLM can figure out the rest.请根据以上内容,在我的 D:/A-原F盘/Obsidian-1/ 仓库里搭建一套相同结构的中文知识库。
● Skill(superpowers:brainstorming) ⎿ Successfully loaded skill
● 好的,让我先探索你当前的 Obsidian 仓库结构,然后我们逐步设计这套系统。
Searched for 2 patterns, read 3 files, listed 2 directories (ctrl+o to expand)
● 好的,我已经了解了你的仓库结构。你有一个 PARA 风格的 Obsidian 仓库,已经在使用 Web Clipper,并且有一个「个人知识库」的初步设置。
现在我有一个关键问题需要先确认:
● 你的仓库目前是 PARA 结构(入口/临时/输入/输出/ToDo)。LLM Wiki 需要自己的三层架构,我倾向于在仓库根目录新建一个独立的 LLM-Wiki/ 顶级文件夹,包含 raw/、wiki/、以及仓库根的 CLAUDE.md,与现有结构平行、互不干扰。
你觉得放在哪个位置更合适?
✻ Brewed for 1m 31s
❯ 按照你的倾向来吧
● 好。让我直接呈现设计方案。
设计方案:LLM Wiki 知识库
一、目录结构
D:/A-原F盘/Obsidian-1/ ├── CLAUDE.md ← Schema:告诉 LLM 如何运作这套 wiki ├── LLM-Wiki/ │ ├── raw/ ← 原始资料(不可变,LLM 只读) │ │ └── .gitkeep │ ├── wiki/ ← LLM 生成的知识库(LLM 全权维护) │ │ ├── index.md ← 内容目录(按分类列出所有页面+简介) │ │ ├── log.md ← 操作日志(按时间追加每次摄入/查询/检查) │ │ ├── _模板/ ← 页面模板(实体页、概念页、摘要页等) │ │ ├── 实体/ ← 人物、组织、工具、产品等 │ │ ├── 概念/ ← 核心概念、方法论、理论 │ │ ├── 摘要/ ← 对 raw/ 中每份资料的摘要 │ │ ├── 对比/ ← 不同事物/观点的比较分析 │ │ └── 综合/ ← 跨资料的综合性论述 │ └── assets/ ← 图片等附件
二、核心文件
- CLAUDE.md:中文 Schema,定义 wiki 结构、页面规范、操作流程(Ingest / Query / Lint)
- index.md:知识目录,LLM 每次摄入后更新,查询时先读它定位相关页面
- log.md:只追加的时间线日志,格式 [YYYY-MM-DD] 操作类型 | 内容描述
三、三种操作
┌──────┬────────────────────────┬────────────────────────────────────────────────────────────────────────┐ │ 操作 │ 触发方式 │ LLM 做什么 │ ├──────┼────────────────────────┼────────────────────────────────────────────────────────────────────────┤ │ 摄入 │ 放入 raw/ 后说「处理」 │ 读源 → 写摘要 → 更新实体/概念页 → 更新 index.md → 追加 log.md │ ├──────┼────────────────────────┼────────────────────────────────────────────────────────────────────────┤ │ 查询 │ 直接提问 │ 读 index.md → 读相关页 → 整合回答(带引用)→ 有价值的内容可回写到 wiki │ ├──────┼────────────────────────┼────────────────────────────────────────────────────────────────────────┤ │ 检查 │ 说「检查 wiki」 │ 扫描矛盾、过时、孤立页、缺失概念 → 给出报告 │ └──────┴────────────────────────┴────────────────────────────────────────────────────────────────────────┘
四、与现有仓库的关系
- 平行于现有的 01入口/ 03输入/ 等目录,互不干扰
- 现有内容可随时作为「源」放入 raw/ 触发摄入
- 仓库已有的 photo/ 附件目录保持不变,raw/ 的图片也可放在 LLM-Wiki/assets/
这个结构是否符合你的预期?有什么想调整的地方吗?
✻ Churned for 32s
❯ 非常好,就这样
● 好,开始创建所有文件。
● Bash(mkdir -p “D:/A-原F盘/Obsidian-1/LLM-Wiki/raw” “D:/A-原F盘/Obsidian-1/LLM-Wiki/wiki/_模板” “D:/A-原F盘/Obsidian-1/LLM-Wiki/wiki/实体” “D:/A-原F盘/Obsidian-1/LLM-Wiki/wiki/概念…) ⎿ Done
────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
Create file
CLAUDE.md
╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌
1 # CLAUDE.md
2
3 本文件为 Claude Code 在此仓库中工作时提供指导。
4
5 仓库概述
6
7 这是一个 Obsidian 知识库,采用 PARA 结构组织(入口/临时/输入/输出和归档/ToD
o)。其中包含一套 LLM Wiki 系统(位于 LLM-Wiki/),由 LLM 增量构建和
维护的结构化知识库。
8
9 三层架构
10
11 第一层:原始资料 LLM-Wiki/raw/
12 - 存放文章的 Markdown 文件、论文、图片、数据文件等
13 - 不可变 — LLM 只能读取,绝不修改原始资料
14 - 由用户负责策展和放入,或通过 Obsidian Web Clipper 剪藏后移入
15 - 图片等附件放在 LLM-Wiki/assets/
16
17 第二层:知识库 LLM-Wiki/wiki/
18 - LLM 全权生成和维护的 Markdown 文件集合
19 - 目录结构:
20 - _模板/ — 页面模板(实体页、概念页、摘要、对比)
21 - 实体/ — 人物、组织、工具、产品、项目等具体事物
22 - 概念/ — 核心概念、方法论、理论、原则
23 - 摘要/ — raw/ 中每份资料的摘要和要点提炼
24 - 对比/ — 不同事物、观点或方法的比较分析
25 - 综合/ — 跨多份资料的综合性论述、主题总览
26 - 规则:LLM 创建、更新、维护所有页面。用户只读。
27
28 第三层:Schema(本文件)
29 - 告诉 LLM wiki 的结构、约定和工作流程
30 - 用户和 LLM 共同演进此文件
31
32 页面规范
33
34 YAML 前置元数据
35 每个 wiki 页面必须在文件顶部包含如下 frontmatter:
36
37 yaml 38 --- 39 title: 页面标题 40 tags: [标签1, 标签2] 41 created: YYYY-MM-DD 42 updated: YYYY-MM-DD 43 type: entity | concept | summary | comparison | synthesis 44 sources: [] # 引用 raw/ 中的源文件 45 --- 46
47
48 命名规范
49 - 文件名 = 页面标题(中文),如 Transformer架构.md
50 - 实体页和概念页的标题应简洁明确
51
52 Wiki 链接
53 - 引用其他 wiki 页面使用 [[页面名]]
54 - 引用原始资料使用 [[../raw/文件名]]
55 - 所有页面之间应建立丰富的交叉引用
56
57 标签体系
58 - 使用扁平标签,如 #LLM #深度学习 #注意力机制
59 - 每个页面 2-5 个标签
60 - 避免过深的层级标签
61
62 操作流程
63
64 操作一:摄入(Ingest)
65
66 当用户说「处理一下」「摄入」「看看这篇」或将文件放入 raw/ 后要求处理时,执
行以下步骤:
67
68 1. 读取源文件:阅读 raw/ 中的新资料
69 2. 讨论要点:与用户讨论关键发现,确认理解方向
70 3. 写入摘要页:在 wiki/摘要/ 创建 {资料名}.md,包含:
71 - 一句话概括
72 - 关键观点(3-5 条)
73 - 新信息/新见解
74 - 与已有知识的关系
75 4. 更新实体/概念页:
76 - 如果资料涉及已有实体或概念,更新对应页面(追加新信息、修正过时内容、标
注矛盾)
77 - 如果引入了新实体或概念,创建对应页面
78 5. 更新交叉引用:确保新旧页面之间的 [[链接]] 完整
79 6. 更新 index.md:在对应分类下添加新页面的链接和一句话描述
80 7. 追加 log.md:记录本次摄入操作
81
82 操作二:查询(Query)
83
84 当用户提问时:
85
86 1. 先读 index.md:定位与问题相关的页面
87 2. 深入阅读:打开相关的 wiki 页面和原始资料
88 3. 整合回答:综合多页面信息,给出带 [[引用]] 的回答
89 4. 有价值的回答回写:如果用户的问题引出了新的分析、对比或连接,主动建议
:
90 > 「这个分析是否值得保存到 wiki 中?」
91 用户可以决定将其写入 对比/ 或 综合/ 目录
92
93 操作三:检查(Lint)
94
95 当用户说「检查一下」「wiki 健康检查」「lint」时,系统性地扫描:
96
97 1. 矛盾检查:不同页面之间是否存在互相矛盾的断言?
98 2. 过时信息:哪些页面的 updated 时间较旧,可能需要用新资料刷新?
99 3. 孤立页面:哪些页面没有被任何其他页面通过 [[链接]] 引用?
100 4. 缺失概念:哪些被多次提及的重要概念还没有自己的独立页面?
101 5. 缺失交叉引用:哪些页面之间明显有关联但没有建立链接?
102 6. 数据缺口:哪些问题可以通过搜索网络补充?
103
104 检查完毕后,给出分级报告:
105 - 🔴 需要立即处理
106 - 🟡 建议改进
107 - 🟢 状态良好
108
109 同时建议 2-3 个值得进一步探索的新问题。
110
111 index.md 约定
112
113 - 按 wiki 子目录分节:实体、概念、摘要、对比、综合
114 - 每行一条:- [[页面名]] — 一句话描述
115 - 摘要节额外标注对应的源文件
116 - 每次摄入后更新
117 - 查询时首先读取此文件
118
119 log.md 约定
120
121 - 只追加,不修改已有条目
122 - 每条格式:## [YYYY-MM-DD] 操作类型 | 内容描述
123 - 操作类型:摄入 查询 检查 维护
117 - 查询时首先读取此文件
118
- 平行于现有的 01入口/ 03输入/ 等目录,互不干扰
- 现有内容可随时作为「源」放入 raw/ 触发摄入
- 仓库已有的 photo/ 附件目录保持不变,raw/ 的图片也可放在 LLM-Wiki/assets/
这个结构是否符合你的预期?有什么想调整的地方吗?
✻ Churned for 32s
❯ 非常好,就这样
● 好,开始创建所有文件。
● Bash(mkdir -p “D:/A-原F盘/Obsidian-1/LLM-Wiki/raw” “D:/A-原F盘/Obsidian-1/LLM-Wiki/wiki/_模板” “D:/A-原F盘/Obsidian-1/LLM-Wiki/wiki/实体” “D:/A-原F盘/Obsidian-1/LLM-Wiki/wiki/概念…) ⎿ Done
● Write(CLAUDE.md)
⎿ Wrote 139 lines to CLAUDE.md
1 # CLAUDE.md
2
3 本文件为 Claude Code 在此仓库中工作时提供指导。
4
5 仓库概述
6
7 这是一个 Obsidian 知识库,采用 PARA 结构组织(入口/临时/输入/输出和归档/ToDo)。其中包含一套 LLM Wiki
系统(位于 LLM-Wiki/),由 LLM 增量构建和维护的结构化知识库。
8
9 三层架构
… +131 lines (ctrl+o to expand)
● 现在并行创建所有 wiki 核心文件。
● Write(LLM-Wiki\wiki\index.md)
● Write(LLM-Wiki\wiki\log.md)
● Write(LLM-Wiki\wiki_模板\实体页模板.md)
● Write(LLM-Wiki\wiki_模板\概念页模板.md)
● Write(LLM-Wiki\wiki_模板\摘要模板.md)
────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
Create file
LLM-Wiki\wiki\index.md
╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌
1 # 知识库目录
2
3 > 本文件由 LLM 维护,每次摄入后更新。查询时首先阅读此文件以定位相关页面。
4
5 ---
6
7 实体
8
9
10
11 暂无内容,等待首次摄入。
12
13 ---
14
15 概念
16
17
18
19 暂无内容,等待首次摄入。
20
21 ---
20
21 ---
7 这是一个 Obsidian 知识库,采用 PARA 结构组织(入口/临时/输入/输出和归档/ToDo)。其中包含一套 LLM Wiki
系统(位于 LLM-Wiki/),由 LLM 增量构建和维护的结构化知识库。
8
9 三层架构
… +131 lines (ctrl+o to expand)
● 现在并行创建所有 wiki 核心文件。
● Write(LLM-Wiki\wiki\index.md)
● Write(LLM-Wiki\wiki\log.md)
● Write(LLM-Wiki\wiki_模板\实体页模板.md)
5 ---
6
7 ## 实体
8
9 <!-- LLM:在此列出人物、组织、工具、产品、项目等实体页面 -->
… +33 lines (ctrl+o to expand)
● Write(LLM-Wiki\wiki\log.md)
● Write(LLM-Wiki\wiki_模板\实体页模板.md)
● Write(LLM-Wiki\wiki_模板\概念页模板.md)
● Write(LLM-Wiki\wiki_模板\摘要模板.md)
● Write(LLM-Wiki\wiki_模板\对比模板.md)
────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
Create file
LLM-Wiki\wiki_模板\实体页模板.md
╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌
1 ---
2 title: {{title}}
3 tags: []
4 created: {{date}}
5 updated: {{date}}
6 type: entity
7 sources: []
8 ---
9
10 # {{title}}
11
12 基本信息
1 ---
2 title: {{title}}
3 tags: []
4 created: {{date}}
5 updated: {{date}}
6 type: entity
7 sources: []
8 ---
9
10 # {{title}}
11
12 基本信息
1 # 操作日志
2
3 > 只追加,不修改。记录 wiki 的演化时间线。
4
5 ---
6
7 [2026-05-10] 初始化 | LLM Wiki 系统搭建
8
9 - 创建了三层架构:raw/ wiki/ CLAUDE.md
10 - 建立了 wiki 子目录:实体、概念、摘要、对比、综合
… +2 lines (ctrl+o to expand)
● Write(LLM-Wiki\wiki_模板\实体页模板.md)
● Write(LLM-Wiki\wiki_模板\概念页模板.md)
● Write(LLM-Wiki\wiki_模板\摘要模板.md)
● Write(LLM-Wiki\wiki_模板\对比模板.md)
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────── Create file LLM-Wiki\wiki_模板\概念页模板.md ╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌ 1 --- 2 title: {{title}} 3 tags: [] 4 created: {{date}} 5 updated: {{date}} 6 type: concept 7 sources: [] 8 --- 9 10 # {{title}} 11 12 一句话定义 13 14 3 tags: [] 4 created: {{date}} 5 updated: {{date}} 6 type: concept 7 sources: [] 8 --- 9 10 # {{title}} 11 12 一句话定义 13 14 ● Write(LLM-Wiki\wiki_模板\实体页模板.md) ⎿ Wrote 34 lines to LLM-Wiki\wiki_模板\实体页模板.md 1 --- 2 title: {{title}} 3 tags: [] 4 created: {{date}} 5 updated: {{date}} 6 type: entity 7 sources: [] 8 --- 9 10 # {{title}} … +24 lines (ctrl+o to expand)
● Write(LLM-Wiki\wiki_模板\概念页模板.md)
● Write(LLM-Wiki\wiki_模板\摘要模板.md)
● Write(LLM-Wiki\wiki_模板\对比模板.md)
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────── Create file LLM-Wiki\wiki_模板\摘要模板.md ╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌ 1 --- 2 title: {{title}} 3 tags: [摘要] 4 created: {{date}} 5 updated: {{date}} 6 type: summary 7 source: {{source_file}} 8 --- 9 10 # {{title}} 11 12 一句话概括 13 14 15 4 created: {{date}} 5 updated: {{date}} 6 type: summary 7 source: {{source_file}} 8 --- 9 10 # {{title}} 11 12 一句话概括 13 14 15 … +24 lines (ctrl+o to expand)
● Write(LLM-Wiki\wiki_模板\概念页模板.md) ⎿ Wrote 36 lines to LLM-Wiki\wiki_模板\概念页模板.md 1 --- 2 title: {{title}} 3 tags: [] 4 created: {{date}} 5 updated: {{date}} 6 type: concept 7 sources: [] 8 --- 9 10 # {{title}} … +26 lines (ctrl+o to expand)
● Write(LLM-Wiki\wiki_模板\摘要模板.md)
● Write(LLM-Wiki\wiki_模板\对比模板.md)
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────── Create file LLM-Wiki\wiki_模板\对比模板.md ╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌ 1 --- 2 title: {{title}} 3 tags: [对比] 4 created: {{date}} 5 updated: {{date}} 6 type: comparison 7 sources: [] 8 --- 9 10 # {{title}} 11 12 对比对象 13 14 - A:实体A或概念A 15 - B:实体B或概念B 16 17 对比维度 18 19 | 维度 | A | B | 8 --- 9 10 # {{title}} 11 12 对比对象 13 14 - A:实体A或概念A 15 - B:实体B或概念B 16 17 对比维度 18 19 | 维度 | A | B | 9 10 # {{title}} … +26 lines (ctrl+o to expand)
● Write(LLM-Wiki\wiki_模板\摘要模板.md) ⎿ Wrote 37 lines to LLM-Wiki\wiki_模板\摘要模板.md 1 --- 2 title: {{title}} 3 tags: [摘要] 4 created: {{date}} 5 updated: {{date}} 6 type: summary 7 source: {{source_file}} 8 --- 9 10 # {{title}} … +27 lines (ctrl+o to expand)
● Write(LLM-Wiki\wiki_模板\对比模板.md) ⎿ Wrote 41 lines to LLM-Wiki\wiki_模板\对比模板.md 1 --- 2 title: {{title}} 3 tags: [对比] 4 created: {{date}} 5 updated: {{date}} 6 type: comparison 7 sources: [] 8 --- 9 10 # {{title}} … +31 lines (ctrl+o to expand)
● 最后创建综合模板,然后验证完整结构。
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────── Create file LLM-Wiki\wiki_模板\综合模板.md ╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌ 1 --- 2 title: {{title}} 3 tags: [综合] 4 created: {{date}} 5 updated: {{date}} 6 type: synthesis 7 sources: [] 8 --- 9 10 # {{title}} 11 12 核心命题 13 14 15 16 背景 17 18 19 20 论述 19 20 论述 … +27 lines (ctrl+o to expand)
● Write(LLM-Wiki\wiki_模板\对比模板.md) ⎿ Wrote 41 lines to LLM-Wiki\wiki_模板\对比模板.md 1 --- 2 title: {{title}} 3 tags: [对比] 4 created: {{date}} 5 updated: {{date}} 6 type: comparison 7 sources: [] 8 --- 9 10 # {{title}} … +31 lines (ctrl+o to expand)
● 最后创建综合模板,然后验证完整结构。
● Write(LLM-Wiki\wiki_模板\综合模板.md) ⎿ Wrote 42 lines to LLM-Wiki\wiki_模板\综合模板.md 1 --- 2 title: {{title}} 3 tags: [综合] 4 created: {{date}} 5 updated: {{date}} 6 type: synthesis 7 sources: [] 8 --- 9 10 # {{title}} … +32 lines (ctrl+o to expand)
Searched for 1 pattern, listed 1 directory (ctrl+o to expand)
● 搭建完成。最终结构如下:
Obsidian-1/ ├── CLAUDE.md ← Schema(操作规范) └── LLM-Wiki/ ├── raw/ ← 放原始资料(你负责) ├── assets/ ← 图片等附件 └── wiki/ ← LLM 主导维护 ├── index.md ← 内容目录 ├── log.md ← 操作日志 ├── _模板/ │ ├── 实体页模板.md │ ├── 概念页模板.md │ ├── 摘要模板.md │ ├── 对比模板.md │ └── 综合模板.md ├── 实体/ ├── 概念/ ├── 摘要/ ├── 对比/ └── 综合/
你现在可以这样使用
- 加入资料:用 Obsidian Web Clipper 剪藏一篇文章,把生成的 .md 拖到 LLM-Wiki/raw/,然后对我说「处理一下」
- 提问查询:直接问我 LLM Wiki 中的内容,我会先读 index.md 定位,再整合回答
- 定期检查:说「检查一下 wiki」,我会扫描矛盾、过时、孤立页面等问题
关键文件说明
- CLAUDE.md(根目录):每次对话自动加载,告诉我 wiki 的结构和操作规则
- index.md:知识库地图,每次摄入后我会自动更新
- log.md:时间线记录,每次操作我都会追加一条
想现在放一篇资料进去试试吗?
✻ Brewed for 3m 54s
※ recap: 已为你搭建了 LLM Wiki 知识库系统(三层架构:raw/、wiki/、CLAUDE.md)。放一篇资料到 LLM-Wiki/raw/
后说「处理一下」即可开始使用。 (disable recaps in /config)
❯ 处理一下
Read 1 file, listed 1 directory (ctrl+o to expand)
● 有一份资料在 raw/ 里,我来读取并处理。
Read 1 file (ctrl+o to expand)
● 摄入资料:孩子坚持去韩国留学
我已读完这篇知乎回答。作者是一位在韩国生活 20 年的人,核心观点是:
韩国留学的现实画像:
- 韩国大学含金量不如国内 985,去的学生多半是国内升学不理想、英语也不顶尖的”半吊子”群体
韩国虽用韩语,但校园里英语决定竞争力,英语好能甩掉半数韩国人;只会中文+韩语半吊子的学生容易困在中国留学生圈子里被歧视
但韩国也有独特优势:
- 地理位置近、生活便利、治安好(亚洲前列)
- 社会制度成熟,追求公平公正人权
- 很多女生留学后选择留在韩国
判断标准:英语好 → 值得去(性价比高);英语一般 → 体验会差。
在读资料的过程中,我注意到这涉及几个可以写成 wiki 页面的概念。在正式写入之前想确认一下:
这个 wiki 你打算聚焦什么主题?目前有无特定的领域方向(比如留学、教育、个人成长),还是先泛化地积累,什么都放?
✻ Worked for 33s
❯ 留学
● 好的,聚焦「留学」领域。现在执行摄入流程——创建摘要页、实体页,然后更新 index.md 和 log.md。
● Write(LLM-Wiki\wiki\摘要\孩子坚持去韩国留学怎么办.md)
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────── Create file LLM-Wiki\wiki\摘要\孩子坚持去韩国留学怎么办.md ╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌ 1 --- 2 title: 孩子坚持去韩国留学怎么办? 3 tags: [留学, 韩国, 知乎问答] 4 created: 2026-05-10 5 updated: 2026-05-10 6 type: summary 7 source: 请问孩子坚持要去韩国留学,怎么办_ - 大树洞 的回答.md 8 --- 9 10 # 孩子坚持去韩国留学怎么办? 11 12 一句话概括 13 14 韩国留学适合英语好、追求性价比和安全的学生,但对语言能力不足、学术水平一般的 学生来说体验可能较差——容易困在中国留学生圈子里,且学历含金量不如国内 985。 15 16 关键观点 17 18 1. 韩国大学含金量有限:不如国内 985 高校,留学生群体整体水平”半吊子”——国 内升学不理想、英语不顶尖、家庭条件一般 19 2. 英语是决定性变量:韩国虽是韩语国家,但校园里英语决定竞争力;英语扎实 可甩掉半数以上韩国人;只会中文+韩语半吊子的学生容易困在中国留学生圈子,饱受 歧视 20 3. 韩国留学的独特优势:地理位置近、生活便利、治安亚洲前列、社会制度成熟 追求公平公正人权、女生留学后留韩率高 21 4. 决策框架:英语好+韩语愿意学 → 性价比高,值得去;英语一般 → 大概率体验 差,不建议 20 3. 韩国留学的独特优势:地理位置近、生活便利、治安亚洲前列、社会制度成熟追求公平公正人权、女生留学后留韩率高 21 4. 决策框架:英语好+韩语愿意学 → 性价比高,值得去;英语一般 → 大概率体验差,不建议 但韩国也有独特优势:
- 地理位置近、生活便利、治安好(亚洲前列)
- 社会制度成熟,追求公平公正人权
- 很多女生留学后选择留在韩国
判断标准:英语好 → 值得去(性价比高);英语一般 → 体验会差。
在读资料的过程中,我注意到这涉及几个可以写成 wiki 页面的概念。在正式写入之前想确认一下:
这个 wiki 你打算聚焦什么主题?目前有无特定的领域方向(比如留学、教育、个人成长),还是先泛化地积累,什么都放?
✻ Worked for 33s
❯ 留学
● 好的,聚焦「留学」领域。现在执行摄入流程——创建摘要页、实体页,然后更新 index.md 和 log.md。
● Write(LLM-Wiki\wiki\摘要\孩子坚持去韩国留学怎么办.md)
● Write(LLM-Wiki\wiki\实体\韩国留学.md)
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────── Create file LLM-Wiki\wiki\实体\韩国留学.md ╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌ 1 --- 2 title: 韩国留学 3 tags: [留学, 韩国, 亚洲留学] 4 created: 2026-05-10 5 updated: 2026-05-10 6 type: entity 7 sources: [孩子坚持去韩国留学怎么办?] 8 --- 9 10 # 韩国留学 11 12 基本信息 13 14 - 目的地:韩国 15 - 语言:韩语(校园英语化程度高) 16 - 地理优势:与中国地理位置近、生活便利 17 - 学位含金量:整体不如国内 985,顶尖校(SKY)可能例外 18 19 概述 20 21 韩国是亚洲热门留学目的地之一,以性价比高、治安好、地理位置近为主要卖点。但其 留学群体整体水平参差不齐,学历含金量存在争议。英语能力是决定留学体验和竞争力 的关键变量。 22 23 关键特征 24 25 - 英语决定体验:韩国高校校园文化偏”美系”,英语扎实的学生能获得显著竞争优 势;语言能力不足则容易被限制在中国留学生圈层内 26 - 治安与社会环境:治安亚洲前列,社会制度成熟,追求公平公正人权,对女性留 学生友好 27 - 性价比定位:相比英美加澳费用较低,适合预算有限但希望获得海外学历的学生 28 - 留韩率:女生留学后选择留在韩国的比例较高 29 - 学生画像:群体中相当比例是国内升学不理想、英语不顶尖、家庭条件一般的学 生 30 31 适合人群 32 33 - 英语口语和学术英语能力强 34 - 愿意学习韩语(至少达到日常生活水平) 35 - 追求留学性价比 24 25 - 英语决定体验:韩国高校校园文化偏”美系”,英语扎实的学生能获得显著竞争优势;语言能力不足则容易被限制在中国留 学生圈层内 26 - 治安与社会环境:治安亚洲前列,社会制度成熟,追求公平公正人权,对女性留学生友好 27 - 性价比定位:相比英美加澳费用较低,适合预算有限但希望获得海外学历的学生 28 - 留韩率:女生留学后选择留在韩国的比例较高 29 - 学生画像:群体中相当比例是国内升学不理想、英语不顶尖、家庭条件一般的学生 30 31 适合人群 32 33 - 英语口语和学术英语能力强 34 - 愿意学习韩语(至少达到日常生活水平) ✻ Worked for 33s
❯ 留学
● 好的,聚焦「留学」领域。现在执行摄入流程——创建摘要页、实体页,然后更新 index.md 和 log.md。
● Write(LLM-Wiki\wiki\摘要\孩子坚持去韩国留学怎么办.md) ⎿ Wrote 43 lines to LLM-Wiki\wiki\摘要\孩子坚持去韩国留学怎么办.md 1 --- 2 title: 孩子坚持去韩国留学怎么办? 3 tags: [留学, 韩国, 知乎问答] 4 created: 2026-05-10 5 updated: 2026-05-10 6 type: summary 7 source: 请问孩子坚持要去韩国留学,怎么办_ - 大树洞 的回答.md 8 --- 9 10 # 孩子坚持去韩国留学怎么办? … +37 lines (ctrl+o to expand)
● Write(LLM-Wiki\wiki\实体\韩国留学.md) ⎿ Wrote 56 lines to LLM-Wiki\wiki\实体\韩国留学.md 1 --- 2 title: 韩国留学 3 tags: [留学, 韩国, 亚洲留学] 4 created: 2026-05-10 5 updated: 2026-05-10 6 type: entity 7 sources: [孩子坚持去韩国留学怎么办?] 8 --- 9 10 # 韩国留学 … +48 lines (ctrl+o to expand)
● 现在更新 index.md 和 log.md。
Read 1 file (ctrl+o to expand)
● Update(LLM-Wiki\wiki\index.md) ⎿ Added 1 line, removed 3 lines 6 7 实体 8 9 - 9 +- 韩国留学 — 韩国作为留学目的地的全面画像:优势、劣势、适合人群与不适合人群 10 11 -暂无内容,等待首次摄入。 12 - 11 --- 12 13 概念
● Update(LLM-Wiki\wiki\index.md)
⎿ Added 1 line, removed 3 lines
20
21 摘要
22
23 -
23 +- 孩子坚持去韩国留学怎么办? — 知乎问答:在韩 20 年答主分析韩国留学的真实画像、英语的关键作用、以及
+什么人适合/不适合去韩国留学 → raw/请问孩子坚持要去韩国留学,怎么办_ - 大树洞 的回答.md
24
25 -暂无内容,等待首次摄入。
26 -
25 ---
26
27 对比
● Update(LLM-Wiki\wiki\log.md) ⎿ Added 8 lines 10 - 建立了 wiki 子目录:实体、概念、摘要、对比、综合 11 - 编写了中文 CLAUDE.md 作为操作规范 12 - 准备好了页面模板和初始 index.md 13 + 14 +## [2026-05-10] 摄入 | 孩子坚持去韩国留学怎么办? 15 + 16 +- 阅读知乎问答(答主在韩 20 年),提取韩国留学真实画像 17 +- 创建摘要页 孩子坚持去韩国留学怎么办? 18 +- 创建实体页 韩国留学(含适合/不适合人群、关键特征) 19 +- 更新 index.md(实体节和摘要节) 20 +- 确定 wiki 聚焦领域:留学
● 摄入完成。以下是本次操作的摘要:
本次摄入:孩子坚持去韩国留学怎么办?
┌───────────────┬─────────────────────────────────────────────────────────────────────────────┐ │ 步骤 │ 结果 │ ├───────────────┼─────────────────────────────────────────────────────────────────────────────┤ │ 源文件 │ raw/ 中的知乎问答(答主在韩 20 年) │ ├───────────────┼─────────────────────────────────────────────────────────────────────────────┤ │ 新建摘要页 │ wiki/摘要/孩子坚持去韩国留学怎么办?.md │ ├───────────────┼─────────────────────────────────────────────────────────────────────────────┤ │ 新建实体页 │ wiki/实体/韩国留学.md — 含适合/不适合人群、关键特征、与其他留学目的地的关联 │ ├───────────────┼─────────────────────────────────────────────────────────────────────────────┤ │ 更新 index.md │ 实体节 + 摘要节已更新 │ ├───────────────┼─────────────────────────────────────────────────────────────────────────────┤ │ 追加 log.md │ 本次操作已记录 │ └───────────────┴─────────────────────────────────────────────────────────────────────────────┘
当前 wiki 状态
wiki/ ├── index.md ← 已更新(1 实体 + 1 摘要) ├── log.md ← 已更新(2 条记录) ├── 实体/ │ └── 韩国留学.md ← 新建 └── 摘要/ └── 孩子坚持去韩国留学怎么办?.md ← 新建
后续你可以继续往 raw/ 里放留学相关的文章、知乎回答、论坛帖子,我会自动关联、对比、补充。放好告诉我就行。
✻ Churned for 1m 38s
❯ 我现在有一个疑问,就是我原来的笔记,也就是输入输出里面的那些,怎样才能这样产出
● 你的现有笔记(03输入/ 04输出和归档/)和 LLM Wiki 的产出之间,差的就是一次「摄入」流程。本质上就是把你自己当作「资料来源」,让我来帮你整理。
具体做法
第一步:挑出来。 从你现有的笔记中,把跟留学相关的挑出来——比如那些关于韩国、留学、语言学习的笔记。
第二步:放进去。 把它们复制或移动到 LLM-Wiki/raw/。
第三步:告诉我。 说「处理一下」,我就会像刚才那样——读你的笔记 → 提炼观点 → 写摘要 → 创建/更新实体页和概念页 → 建立交叉引用。
举个例子
你的 03输入/学习/韩语.md 如果放进 raw/,我摄入后会:
- 创建 摘要/韩语学习笔记.md
- 更新 实体/韩国留学.md,补充「语言准备」相关内容
- 可能创建 概念/留学语言准备.md
两种模式的区别
┌──────┬───────────────────┬──────────────────────────────────┐ │ │ 你原来的笔记 │ LLM Wiki 产出 │ ├──────┼───────────────────┼──────────────────────────────────┤ │ 谁写 │ 你手写 │ LLM 生成 │ ├──────┼───────────────────┼──────────────────────────────────┤ │ 格式 │ 自由,因人而异 │ 统一 frontmatter + 结构 │ ├──────┼───────────────────┼──────────────────────────────────┤ │ 关联 │ 靠你手动 链接 │ LLM 自动建立交叉引用 │ ├──────┼───────────────────┼──────────────────────────────────┤ │ 更新 │ 写完就忘 │ 每次新资料摄入时自动刷新相关页面 │ ├──────┼───────────────────┼──────────────────────────────────┤ │ 形态 │ 碎片化的散点 │ 结构化的知识网络 │ └──────┴───────────────────┴──────────────────────────────────┘
你原来的笔记是「素材」,LLM Wiki 是「加工后的知识」。你不必把所有笔记都放进 wiki——只放那些你想让 LLM 帮你提炼、串联、持续积累的。
想试试把某个现有笔记放进来处理吗?
✻ Cogitated for 24s
❯ 明白了
● 好的,随时往 raw/ 里放资料然后说「处理一下」就行。
✻ Cogitated for 3s
相关链接
Obsidian + Claude Code 完整真实过程 | Obsidian Web Clipper | Obsidian 加 Hermes 加 Cloud Code | Obsidian 加 Hermes 加 Cloud Code中hermes的意义 | Obsidian 加 Hermes 加 Cloud Code加Copilot Settings插件,能组合出什么惊艳的知识库效果吗