What is Karpathy's LLM Wiki?

Andrej Karpathy's pattern for LLM-maintained markdown wikis, beating RAG by persisting knowledge incrementally.

Hjarni vs local LLM Wiki: Which to choose?

Local for git lovers and single-machine purists. Hjarni for cross-device, multi-LLM access without sync pain.

Does LLM Wiki replace RAG entirely?

For personal or team knowledge bases, yes – compounds value. RAG stays for massive, unstructured doc search.

Karpathy's LLM Wiki Nails It – But Local Setup's Friction Killed It for Me

What if your AI notes actually remembered everything, without rehashing docs every query? Karpathy's LLM Wiki does that – I built Hjarni to fix its biggest pains.

theAIcatchup Apr 09, 2026 4 min read

Andrej Karpathy's LLM Wiki setup in Obsidian with Claude Code integration

⚡ Key Takeaways

Karpathy's LLM Wiki fixes RAG's rediscovery waste with persistent, LLM-maintained markdown knowledge. 𝕏
Local setups suffer friction: single-machine, single-client, hard sharing – killing habits. 𝕏
Hjarni hosts the pattern over MCP for smoothly multi-device, multi-LLM access, trading git for ubiquity. 𝕏

Published by

theAIcatchup

Ship faster. Build smarter.

#Andrej Karpathy #Hjarni #LLM Wiki #RAG alternative

Worth sharing?

Get the best Developer Tools stories of the week in your inbox — no noise, no spam.

Originally reported by dev.to

⚡ Key Takeaways

The 60-Second TL;DR

theAIcatchup

Share this article

Worth sharing?

Related Stories

Fake Token Hijacks Solana's Drift Governance — $285M Gone in 12 Minutes

AWS Cells: The Hidden Trick Scaling S3 to Trillions Without Exploding

Meta's Muse Spark: 16 Tools That Mock Every Other AI Chatbot

Claude Agents That Won't Implode: The Three Production Layers Devs Ignore

Stay in the loop