llm-wiki

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

LLM Wiki — Second Brain for Claude Code + Obsidian

LLM Wiki — 适用于Claude Code + Obsidian的第二大脑

Inspired by Andrej Karpathy's LLM Wiki pattern (gist). This skill turns Claude Code (or any agent CLI) into a disciplined wiki maintainer that incrementally builds and maintains a persistent, interlinked Obsidian vault as you feed it sources. The knowledge compounds — cross-references, contradictions, and synthesis are already there when you query.

灵感来自Andrej Karpathy的LLM Wiki模式（gist）。该技能可将Claude Code（或任何Agent CLI）转变为严谨的维基维护者，在你导入数据源时逐步构建并维护一个持久化、相互关联的Obsidian知识库。知识会不断复利——当你查询时，交叉引用、矛盾点和整合内容已准备就绪。

Core principle

核心原则

Most LLM+docs workflows are RAG: retrieve fragments at query time, synthesize from scratch, forget. The wiki is compounding: sources are read once, integrated into a persistent markdown knowledge base, and kept current. You curate and ask; the LLM reads, files, cross-references, and maintains.

Obsidian is the IDE. The LLM is the programmer. The wiki is the codebase.

大多数LLM+文档工作流采用RAG：查询时检索片段，从头整合，用完即忘。而本维基采用复利模式：数据源仅读取一次，整合到持久化Markdown知识库中并保持时效性。你负责策划和提问；LLM负责阅读、归档、交叉引用和维护。

Obsidian是集成开发环境（IDE），LLM是程序员，维基是代码库。

When to use

适用场景

Personal: track goals, health, psychology, journaling, self-improvement
Research: deep dives over weeks on a topic — papers, articles, reports, evolving thesis
Book companion: file chapters as you read; build a fan-wiki-style companion for characters, themes, plot threads
Business/team: internal wiki fed by Slack, meeting notes, calls — LLM does maintenance nobody else wants to do
Competitive analysis, due diligence, trip planning, course notes, hobby deep-dives

Do NOT use when: you need one-shot Q&A over a fixed document (use RAG), you don't plan to add sources over time, or you don't want Obsidian in the loop.

个人场景：追踪目标、健康、心理状态、日志记录、自我提升
研究场景：针对某一主题进行数周深度研究——论文、文章、报告、不断演进的论点
书籍辅助：阅读时归档章节；构建类似粉丝维基的内容，涵盖角色、主题、情节线
商务/团队场景：由Slack消息、会议记录、通话内容填充的内部维基——LLM承担无人愿意做的维护工作
竞品分析、尽职调查、旅行规划、课程笔记、爱好深度研究

不适用场景：你需要对固定文档进行一次性问答（使用RAG即可）、不打算随时间添加数据源，或不想让Obsidian参与流程。

Architecture (three layers)

架构（三层）

vault/
├── raw/                    # Layer 1 — IMMUTABLE source of truth
│   ├── <source files>      # Articles, papers, PDFs, images, data
│   └── assets/             # Downloaded images from clipped articles
├── wiki/                   # Layer 2 — LLM-owned knowledge base
│   ├── index.md            # Content catalog (LLM updates every ingest)
│   ├── log.md              # Append-only timeline (## [YYYY-MM-DD] <op> | <title>)
│   ├── entities/           # Person/Org/Place pages
│   ├── concepts/           # Ideas, theories, frameworks
│   ├── sources/            # One summary page per ingested source
│   ├── comparisons/        # Cross-source analysis pages
│   └── synthesis/          # High-level syntheses, theses, overviews
├── CLAUDE.md               # Schema + conventions (Claude Code)
└── AGENTS.md               # Same content, for Codex/Cursor/Antigravity

Layer 1 (raw/) — you own. LLM only reads; never writes.
Layer 2 (wiki/) — LLM owns. It creates, updates, and cross-references pages. You read it.
Layer 3 (CLAUDE.md / AGENTS.md) — the schema. Conventions, workflows, frontmatter rules. Co-evolved by you and the LLM.

vault/
├── raw/                    # 第一层 — 不可变的事实来源
│   ├── <source files>      # 文章、论文、PDF、图片、数据
│   └── assets/             # 从截取文章中下载的图片
├── wiki/                   # 第二层 — LLM专属知识库
│   ├── index.md            # 内容目录（LLM每次导入时更新）
│   ├── log.md              # 仅追加的时间线（## [YYYY-MM-DD] <操作> | <标题>）
│   ├── entities/           # 人物/组织/地点页面
│   ├── concepts/           # 观点、理论、框架
│   ├── sources/            # 每个导入数据源对应的摘要页面
│   ├── comparisons/        # 跨数据源分析页面
│   └── synthesis/          # 高层级整合内容、论点、概述
├── CLAUDE.md               #  Schema + 约定规则（供Claude Code使用）
└── AGENTS.md               # 相同内容，供Codex/Cursor/Antigravity使用

第一层（raw/） — 由你掌控。LLM仅读取，绝不写入。
第二层（wiki/） — 由LLM掌控。它负责创建、更新和交叉引用页面。你可以阅读这些内容。
第三层（CLAUDE.md / AGENTS.md） — Schema。包含约定规则、工作流、前置元数据规则。由你和LLM共同演进。

Three core operations

三项核心操作

Ingest — LLM reads a source, discusses takeaways with you, writes a source summary, updates 10-15 relevant pages, updates index, appends to log. See
```
references/ingest-workflow.md
```
.
Query — LLM reads
```
index.md
```
first, drills into relevant pages, synthesizes with citations. Good answers get filed back into the wiki so explorations compound. See
```
references/query-workflow.md
```
.
Lint — Health check: contradictions, stale claims, orphan pages, missing cross-refs, concepts mentioned but lacking their own page, data gaps to fill with web search. See
```
references/lint-workflow.md
```
.

导入（Ingest） — LLM读取数据源，与你讨论要点，撰写数据源摘要，更新10-15个相关页面，更新目录，追加到日志。详见
```
references/ingest-workflow.md
```
。
查询（Query） — LLM先读取
```
index.md
```
，深入相关页面，结合引用内容生成整合答案。优质答案会重新归档到维基，让探索内容持续复利。详见
```
references/query-workflow.md
```
。
检查（Lint） — 健康检查：识别矛盾点、过时声明、孤立页面、缺失的交叉引用、被提及但未创建独立页面的概念、需通过网络搜索填补的数据缺口。详见
```
references/lint-workflow.md
```
。

Quick start

快速开始

bash

undefined

bash

undefined

1. Initialize a vault (in Obsidian's vault directory)

1. 初始化知识库（在Obsidian的知识库目录中）

python scripts/init_vault.py --path ~/vaults/research --topic "LLM interpretability"

2. Drop a source into raw/, then ingest

2. 将数据源放入raw/，然后导入

/wiki-ingest ~/vaults/research/raw/anthropic-monosemanticity.pdf

3. Ask questions (answers can be re-filed into the wiki)

3. 提问（答案可重新归档到维基）

/wiki-query "how does monosemanticity compare to mechanistic interpretability?"

4. Periodic health check

4. 定期健康检查

/wiki-lint

5. See the timeline

5. 查看时间线

/wiki-log --last 10

undefined

/wiki-log --last 10

undefined

Slash commands (this plugin ships)

斜杠命令（本插件内置）

Command	Purpose
`/wiki-init`	Bootstrap a fresh vault with schema files + starter structure
`/wiki-ingest <path>`	Read a source, discuss, update wiki, log it
`/wiki-query <question>`	Search wiki, synthesize answer, offer to file back
`/wiki-lint`	Run health check — contradictions, orphans, stale claims, gaps
`/wiki-log`	Show recent log entries (uses unix tools on `log.md` )

命令	用途
`/wiki-init`	用Schema文件和初始结构引导创建新知识库
`/wiki-ingest <路径>`	读取数据源，讨论内容，更新维基，记录日志
`/wiki-query <问题>`	搜索维基，生成整合答案，提供归档选项
`/wiki-lint`	运行健康检查——识别矛盾点、孤立页面、过时声明、缺口
`/wiki-log`	显示近期日志条目（基于 `log.md` 使用Unix工具）

Sub-agents (this plugin ships)

子Agent（本插件内置）

Agent	When dispatched
`wiki-ingestor`	Delegated ingest flow — reads source, proposes updates, applies after your approval
`wiki-linter`	Runs the health-check workflow independently, reports findings
`wiki-librarian`	Answers queries using index-first search, synthesizes with citations

Agent	调度时机
`wiki-ingestor`	委托导入流程——读取数据源，提议更新内容，经你批准后应用
`wiki-linter`	独立运行健康检查工作流，报告发现的问题
`wiki-librarian`	使用目录优先搜索回答问题，结合引用内容生成整合答案

Python tools (

scripts/

)

Python工具（

scripts/

目录）

All tools are standard library only (no pip installs). Run with

python scripts/<tool>.py --help

Script	Purpose
`init_vault.py`	Create folder structure + seed CLAUDE.md, AGENTS.md, index.md, log.md
`ingest_source.py`	Helper: extract text/frontmatter from a source file, ready for LLM review
`update_index.py`	Regenerate `index.md` from wiki page frontmatter (category, date, source count)
`append_log.py`	Append a standardized log entry `## [YYYY-MM-DD] <op> \| <title>`
`wiki_search.py`	BM25 search over wiki pages (standalone fallback when index.md isn't enough)
`lint_wiki.py`	Find orphans (no inbound links), stale pages, missing cross-refs, broken links
`graph_analyzer.py`	Compute link graph stats — hubs, orphans, clusters, disconnected components
`export_marp.py`	Render a wiki page (or subtree) to a Marp slide deck

所有工具仅依赖标准库（无需pip安装）。运行方式：

python scripts/<工具名>.py --help

。

脚本	用途
`init_vault.py`	创建文件夹结构 + 初始化CLAUDE.md、AGENTS.md、index.md、log.md
`ingest_source.py`	辅助工具：从数据源文件提取文本/前置元数据，供LLM审阅
`update_index.py`	根据维基页面的前置元数据（分类、日期、数据源数量）重新生成 `index.md`
`append_log.py`	追加标准化日志条目 `## [YYYY-MM-DD] <操作> \| <标题>`
`wiki_search.py`	对维基页面进行BM25搜索（当index.md不足以满足需求时的独立备选方案）
`lint_wiki.py`	查找孤立页面（无入站链接）、过时页面、缺失的交叉引用、失效链接
`graph_analyzer.py`	计算链接图统计数据——枢纽页面、孤立页面、集群、断开的组件
`export_marp.py`	将维基页面（或子树）渲染为Marp幻灯片

Cross-tool compatibility

跨工具兼容性

The vault's schema lives in CLAUDE.md (Claude Code) or AGENTS.md (Codex/Cursor/Antigravity/OpenCode). The same content works in both. This plugin ships both templates. For per-tool setup instructions see

references/cross-tool-setup.md

CLAUDE.md       → Claude Code
AGENTS.md       → Codex CLI, Cursor, Antigravity, OpenCode, Gemini CLI
.cursorrules    → legacy Cursor (pre-AGENTS.md)

The scripts are pure Python stdlib → run identically everywhere. Only the loader file changes per tool.

知识库的Schema存储在CLAUDE.md（供Claude Code使用）或AGENTS.md（供Codex/Cursor/Antigravity/OpenCode使用）中。两份内容完全通用。本插件内置两种模板。各工具的设置说明详见

references/cross-tool-setup.md

。

CLAUDE.md       → Claude Code
AGENTS.md       → Codex CLI、Cursor、Antigravity、OpenCode、Gemini CLI
.cursorrules    → 旧版Cursor（AGENTS.md之前的版本）

脚本为纯Python标准库实现 → 在所有环境中运行方式一致。仅加载文件会因工具而异。

Obsidian setup (recommended)

Obsidian推荐设置

Obsidian Web Clipper — browser extension; converts web articles to markdown and drops them in
```
raw/
```
Download images locally — Settings → Files and links → Attachment folder path =
```
raw/assets/
```
. Settings → Hotkeys → bind "Download attachments for current file" to
```
Ctrl+Shift+D
```
Graph view — see hubs/orphans; essential for spotting structural problems
Marp plugin — Markdown-based slide decks directly from wiki pages
Dataview plugin — dynamic tables/lists over page frontmatter (tags, dates, source counts)
Git — the vault is a plain markdown repo; version it

Full setup walkthrough:

references/obsidian-setup.md

Obsidian Web Clipper — 浏览器扩展；将网页文章转换为Markdown并放入
```
raw/
```
目录
本地下载图片 — 设置 → 文件与链接 → 附件文件夹路径 =
```
raw/assets/
```
。设置 → 快捷键 → 将“下载当前文件的附件”绑定为
```
Ctrl+Shift+D
```
图谱视图 — 查看枢纽/孤立页面；对发现结构问题至关重要
Marp插件 — 直接从维基页面生成基于Markdown的幻灯片
Dataview插件 — 基于页面前置元数据（标签、日期、数据源数量）生成动态表格/列表
Git — 知识库是纯Markdown仓库；进行版本控制

完整设置指南：

references/obsidian-setup.md

Why this works (vs plain RAG)

为何该方案优于普通RAG

Plain RAG	LLM Wiki
Rediscover knowledge each query	Knowledge accumulates
Cross-references re-computed every time	Cross-references pre-written and maintained
Contradictions surface only if you ask	Contradictions flagged during ingest
Exploration disappears into chat history	Good answers re-filed as new pages
Scales by embeddings infrastructure	Scales by markdown + `index.md` + optional local search

At ~100 sources / hundreds of pages,

index.md

+ filesystem search is enough. Past that, layer in a local search tool like qmd or use

scripts/wiki_search.py

普通RAG	LLM Wiki
每次查询都需重新检索知识	知识持续积累
交叉引用每次查询都需重新计算	交叉引用已预先撰写并维护
仅当你主动提问时才会发现矛盾点	导入时即标记矛盾点
探索内容仅存在于聊天记录中	优质答案会重新归档为新页面
依赖嵌入向量基础设施实现扩展	通过Markdown + `index.md` + 可选本地搜索实现扩展

当数据源达到约100个/页面达到数百个时，

index.md

+ 文件系统搜索已足够。超过该规模后，可添加本地搜索工具如qmd或使用

scripts/wiki_search.py

。

Related skills (chains via

context: fork

)

Reference docs

参考文档

```
references/wiki-schema.md
```
— full vault layout, page frontmatter, naming conventions
```
references/page-formats.md
```
— entity, concept, source, comparison, synthesis templates
```
references/ingest-workflow.md
```
— the detailed ingest flow the wiki-ingestor agent follows
```
references/query-workflow.md
```
— query patterns, citation format, re-filing answers
```
references/lint-workflow.md
```
— health-check heuristics
```
references/obsidian-setup.md
```
— Obsidian plugins, hotkeys, vault config
```
references/cross-tool-setup.md
```
— per-tool setup (Codex, Cursor, Antigravity, etc.)
```
references/memex-principles.md
```
— Bush's Memex, why the LLM changes the maintenance math

```
references/wiki-schema.md
```
— 完整知识库布局、页面前置元数据、命名约定
```
references/page-formats.md
```
— 实体、概念、数据源、对比、整合内容的模板
```
references/ingest-workflow.md
```
— wiki-ingestor Agent遵循的详细导入流程
```
references/query-workflow.md
```
— 查询模式、引用格式、答案重新归档规则
```
references/lint-workflow.md
```
— 健康检查启发式规则
```
references/obsidian-setup.md
```
— Obsidian插件、快捷键、知识库配置
```
references/cross-tool-setup.md
```
— 各工具设置说明（Codex、Cursor、Antigravity等）
```
references/memex-principles.md
```
— Bush的Memex原理，为何LLM改变了维护逻辑

Templates (

assets/

)

模板（

assets/

目录）

CLAUDE.md.template

AGENTS.md.template

.cursorrules.template

— schema loaders per tool

```
index.md.template
```
,
```
log.md.template
```
— starter index and log
```
page-templates/
```
— entity, concept, source-summary, comparison, synthesis
```
example-vault/
```
— small worked example you can study or copy

CLAUDE.md.template

AGENTS.md.template

.cursorrules.template

— 各工具对应的Schema加载器

```
index.md.template
```
,
```
log.md.template
```
— 初始目录和日志模板
```
page-templates/
```
— 实体、概念、数据源摘要、对比、整合内容的模板
```
example-vault/
```
— 可研究或复制的小型示例知识库

Iron rule

铁则

The LLM never edits files in
raw/
. Ever. Sources are immutable. All LLM writes go to

wiki/

. If you need to correct a source, do it in

raw/

yourself — then re-ingest.

LLM绝不编辑
raw/
目录下的文件。永远不会。数据源是不可变的。所有LLM写入操作都仅针对

wiki/

目录。若需修正数据源，请自行在

raw/

目录中修改——然后重新导入。