Karpathy posted about his raw/ folder workflow on April 2. Five days later, Graphify is sitting at #4 on Trendshift with 2,300+ stars.
What It Does
Type /graphify in Claude Code (or Codex, OpenCode, OpenClaw). It reads your entire codebase — code, docs, PDFs, images, whiteboard photos — and builds a persistent knowledge graph. Two-phase approach: first, deterministic AST parsing extracts code structure without touching an LLM. Then Claude subagents run in parallel across documents and images, pulling out concepts and relationships. Everything merges into a NetworkX graph with Leiden community detection, exported as interactive HTML and queryable JSON.
The headline number: 71.5x fewer tokens per query compared to dumping raw files into context. That’s not a rounding error — it’s the difference between burning $7 and burning $0.10.
Why It’s Blowing Up
Karpathy’s post created the demand. Graphify filled it. The .NET community already ported it (graphify-dotnet dropped April 6), which tells you developers aren’t just starring — they’re building on it. The fact that it’s multimodal matters too. Most knowledge graph tools choke on anything that isn’t text. Graphify treats screenshots and diagrams as first-class inputs.
MIT licensed. One command install.
You Might Also Like
- 27k Github Stars in Weeks Learn Claude Code by Shareai lab Breaks Down ai Coding Agents Into 12 Lessons
- Claude hud hit 5 3k Github Stars Because Developers Were Flying Blind With Claude Code
- 27 Agents 109 Skills 88k Github Stars is Everything Claude Code Genius or Over Engineering
- 708 Github Stars in 48 Hours Claude Token Efficient Universal Claude md and the Fight Over Claudes Most Expensive Habit
- Pi Mono 29k Stars and a 200 Token System Prompt That Rivals Claude Code

Leave a comment