Skip to content

LLM Wiki is a cross-platform desktop application that turns your documents into an organized, interlinked knowledge base — automatically. Instead of traditional RAG (retrieve-and-answer from scratch every time), the LLM incrementally builds and maintains a persistent wiki from your sources。

TypeScript

docs: add screenshots, fix Karpathy gist URL, add architecture diagram

nash_su20a03e2b4ed7
diff --git a/README.md b/README.md
index 6f602240..791bc37e 100644
--- a/README.md
+++ b/README.md
@@ -24,15 +24,23 @@
---
+<p align="center">
+ <img src="assets/3-knowledge_graph.jpg" width="100%" alt="Knowledge Graph">
+</p>
+
## What is this?
LLM Wiki is a cross-platform desktop application that turns your documents into an organized, interlinked knowledge base — automatically. Instead of traditional RAG (retrieve-and-answer from scratch every time), the LLM **incrementally builds and maintains a persistent wiki** from your sources. Knowledge is compiled once and kept current, not re-derived on every query.
-This project is based on [Karpathy's LLM Wiki pattern](https://github.com/karpathy/llm-wiki) — a methodology for building personal knowledge bases using LLMs. We implemented the core ideas as a full desktop application with significant enhancements.
+This project is based on [Karpathy's LLM Wiki pattern](https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f) — a methodology for building personal knowledge bases using LLMs. We implemented the core ideas as a full desktop application with significant enhancements.
+
+<p align="center">
+ <img src="assets/llm_wiki_arch.jpg" width="100%" alt="LLM Wiki Architecture">
+</p>
## Credits
-The foundational methodology comes from **Andrej Karpathy**'s [llm-wiki.md](https://github.com/karpathy/llm-wiki), which describes the pattern of using LLMs to incrementally build and maintain a personal wiki. The original document is an abstract design pattern; this project is a concrete implementation with substantial extensions.
+The foundational methodology comes from **Andrej Karpathy**'s [llm-wiki.md](https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f), which describes the pattern of using LLMs to incrementally build and maintain a personal wiki. The original document is an abstract design pattern; this project is a concrete implementation with substantial extensions.
## What We Kept from the Original
@@ -47,6 +55,10 @@ The core architecture follows Karpathy's design faithfully:
- **Obsidian compatibility** — the wiki directory works as an Obsidian vault
- **Human curates, LLM maintains** — the fundamental role division
+<p align="center">
+ <img src="assets/5-obsidian_compatibility.jpg" width="100%" alt="Obsidian Compatibility">
+</p>
+
## What We Changed & Added
### 1. From CLI to Desktop Application
@@ -94,6 +106,10 @@ Additional ingest enhancements beyond the original:
### 4. Knowledge Graph with Relevance Model
+<p align="center">
+ <img src="assets/3-knowledge_graph.jpg" width="100%" alt="Knowledge Graph">
+</p>
+
The original mentions `[[wikilinks]]` for cross-references but has no graph analysis. We built a **full knowledge graph visualization and relevance engine**:
**4-Signal Relevance Model:**
@@ -178,6 +194,10 @@ The original suggests staying involved during ingest. We added an **asynchronous
### 10. Deep Research
+<p align="center">
+ <img src="assets/1-deepresearch.jpg" width="100%" alt="Deep Research">
+</p>
+
Not in the original. When the LLM identifies knowledge gaps:
- **Web search** (Tavily API) finds relevant sources
@@ -189,6 +209,10 @@ Not in the original. When the LLM identifies knowledge gaps:
### 11. Browser Extension (Web Clipper)
+<p align="center">
+ <img src="assets/4-chrome_extension_webclipper.jpg" width="100%" alt="Chrome Extension Web Clipper">
+</p>
+
The original mentions Obsidian Web Clipper. We built a **dedicated Chrome Extension** (Manifest V3):
- **Mozilla Readability.js** for accurate article extraction (strips ads, nav, sidebars)
diff --git a/README_CN.md b/README_CN.md
index b8e8c0c1..8873a41b 100644
--- a/README_CN.md
+++ b/README_CN.md
@@ -24,19 +24,27 @@
---
+<p align="center">
+ <img src="assets/3-knowledge_graph.jpg" width="100%" alt="知识图谱">
+</p>
+
## 这是什么?
LLM Wiki 是一个跨平台桌面应用,能将你的文档自动转化为有组织、相互关联的知识库。与传统 RAG(每次查询都从头检索和回答)不同,LLM 会从你的资料中**增量构建并维护一个持久化的 Wiki**。知识只编译一次并持续更新,而非每次查询都重新推导。
-本项目基于 [Karpathy 的 LLM Wiki 模式](https://github.com/karpathy/llm-wiki) —— 一套使用 LLM 构建个人知识库的方法论。我们将其核心理念实现为一个完整的桌面应用,并做了大量增强。
+本项目基于 [Karpathy 的 LLM Wiki 方法论](https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f) —— 一套使用 LLM 构建个人知识库的方法论。我们将其核心理念实现为一个完整的桌面应用,并做了大量增强。
+
+<p align="center">
+ <img src="assets/llm_wiki_arch.jpg" width="100%" alt="LLM Wiki 架构图">
+</p>
## 致谢
-基础方法论来自 **Andrej Karpathy** 的 [llm-wiki.md](https://github.com/karpathy/llm-wiki),描述了使用 LLM 增量构建和维护个人 Wiki 的设计模式。原始文档是一个抽象的设计范式;本项目是一个具体的实现,并有大量扩展。
+基础方法论来自 **Andrej Karpathy** 的 [llm-wiki.md](https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f),描述了使用 LLM 增量构建和维护个人 Wiki 的设计模式。原始文档是一个抽象的设计范式;本项目是一个具体的实现,并有大量扩展。
## 保留的原始设计
-核心架构忠实遵循 Karpathy 的设计:
+核心架构忠实遵循 Karpathy 的方法论:
- **三层架构**:原始资料(不可变)→ Wiki(LLM 生成)→ Schema(规则和配置)
- **三个核心操作**:Ingest(摄入)、Query(查询)、Lint(检查)
@@ -47,6 +55,10 @@ LLM Wiki 是一个跨平台桌面应用,能将你的文档自动转化为有
- **Obsidian 兼容** —— Wiki 目录可直接作为 Obsidian 仓库使用
- **人类策展,LLM 维护** —— 基本角色分工
+<p align="center">
+ <img src="assets/5-obsidian_compatibility.jpg" width="100%" alt="Obsidian 兼容">
+</p>
+
## 我们的修改与新增
### 1. 从命令行到桌面应用
@@ -94,6 +106,10 @@ LLM Wiki 是一个跨平台桌面应用,能将你的文档自动转化为有
### 4. 知识图谱与关联度模型
+<p align="center">
+ <img src="assets/3-knowledge_graph.jpg" width="100%" alt="知识图谱">
+</p>
+
原始设计提到了 `[[wikilinks]]` 用于交叉引用,但没有图分析。我们构建了**完整的知识图谱可视化和关联度引擎**:
**四信号关联度模型:**
@@ -178,6 +194,10 @@ LLM Wiki 是一个跨平台桌面应用,能将你的文档自动转化为有
### 10. 深度研究
+<p align="center">
+ <img src="assets/1-deepresearch.jpg" width="100%" alt="深度研究">
+</p>
+
原始设计中没有。当 LLM 识别出知识空白时:
- **网络搜索**(Tavily API)查找相关资料
@@ -189,6 +209,10 @@ LLM Wiki 是一个跨平台桌面应用,能将你的文档自动转化为有
### 11. 浏览器扩展(网页剪藏)
+<p align="center">
+ <img src="assets/4-chrome_extension_webclipper.jpg" width="100%" alt="Chrome 扩展网页剪藏">
+</p>
+
原始设计提到了 Obsidian Web Clipper。我们构建了**专用 Chrome 扩展**(Manifest V3):
- **Mozilla Readability.js** 精确提取文章内容(去除广告、导航、侧边栏)