sdyckjq-lab

sdyckjq-lab / llm-wiki-skill

Public

基于 Karpathy llm-wiki 方法论的个人知识库构建 Skill，支持多平台！

316
38
89% credibility
Found Apr 06, 2026 at 120 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Shell
AI Summary

A toolkit for AI agents to automatically extract, organize, and link content from web sources into a personal Markdown-based wiki compatible with note-taking apps.

How It Works

1
🔍 Discover the wiki builder

You hear about a handy tool that turns scattered online info like articles, tweets, and videos into a neat, linked-up personal knowledge collection for your AI helper.

2
🤖 Share with your AI assistant

Simply give the project link to your AI companion like Claude or OpenClaw, and ask it to set things up.

3
🛠️ Assistant prepares your toolkit

Your AI handles the setup automatically, creating folders ready for your knowledge base.

4
🆕 Start your knowledge wiki

Tell your AI to create the wiki structure, and it generates pages, indexes, and templates just for you.

5
Add your materials
🌐
Web articles or posts

Paste a webpage or blog link, and it pulls out the clean text.

🐦
Tweets or threads

Share a Twitter/X link for full conversations.

🎥
Videos

Give a YouTube link to grab the subtitles.

📁
Files or text

Drop in PDFs, notes, or plain text directly.

6
🔗 AI builds linked pages

Your assistant smartly summarizes, connects ideas with links, and organizes into topics, entities, and overviews.

✨ Explore your growing wiki

Open in a note app like Obsidian or ask your AI questions—your knowledge is now structured, searchable, and keeps improving.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 120 to 316 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is llm-wiki-skill?

This Shell-based skill turns scattered web links, X posts, YouTube transcripts, and files into a persistent, linked Markdown wiki for LLM agents like Claude Code, Codex, and OpenClaw. Inspired by Karpathy's llm-wiki method from his GitHub gist—tied to his karpathy github ai research like llm-council and nanogpt—it compiles knowledge once into entities, topics, and summaries, avoiding repeated derivations. Agents get a raw素材 folder feeding a structured wiki, Obsidian-ready with double-bracket links.

Why is it gaining traction?

It stands out with one-line installs via `bash install.sh --platform claude` (or openclaw/codex), auto-routing URLs to extractors for web articles, X threads, and more, plus wiki health checks for orphans and conflicts. Developers dig the zero-config init and batch processing, echoing Karpathy github repo efficiency from makemore to rnn effectiveness studies. At 27 stars, it's niche but hooks agent users tired of prompt bloating.

Who should use this?

AI agent wranglers on Claude Code or OpenClaw building personal code wikis from GitHub karpathy github council threads, research papers, or tool docs. Ideal for devs using Codex for ongoing projects needing linked entity pages over raw pastes. Skip if you're not shell-savvy or lack Chrome for extractions.

Verdict

Grab it if you're deep in Claude or OpenClaw workflows—solid docs and Karpathy roots make setup painless despite 27 stars and 0.9% credibility score signaling early maturity. Test on a small wiki first; scales well for skill-heavy LLM coding.

(187 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.