Fix Karpathy’s LLM Wiki with a Knowledge Graph | Claude Code + Obsidian + InfraNodus

In this video, I'm going to show how Andrej Karpathy's LLM wiki setup can be improved using https://infranodus.com — using knowledge graphs to discover topical clusters and content gaps between your ideas in order to generate better insights. The easiest way to use it is to install the InfraNodus extension to your favorite IDE where you use Claude Code, for instance, for Cursor or Antigravity — https://infranodus.com/vscode-extension You can also use InfraNodus inside Obsidian if you prefer it for visualizing your wiki: https://infranodus.com/obsidian-plugin To set up your own LLM wiki, you can use our Claude skill that will take you through the process: https://github.com/infranodus/skills/blob/master/skill-llm-wiki/SKILL.md Timestamps: 0:00 LLM Wiki — organise your knowledge 1:12 The problem with LLM Wiki 2:12 Real-life demo 3:41 Knowledge graphs help visualise and optimize the structure of the LLM wiki 5:00 Obsidian InfraNodus graph view 5:36 MCP server to create graphs 6:53 Adding knowledge graphs into LLM wiki 8:23 “You generate a living memory of what’s happening in the system” 8:57 DEMO: SETTING UP YOUR LLM WIKI with our skill 13:10 Ingesting sources into LLM Wiki 15:21 Organising ideas into the internal wiki 15:39 Exploring and optimising the content of Wiki with InfraNodus graph 18:05 Using the knowledge graph to bridge a content gap in your wiki 20:22 Adding a plan to your research 22:23 Final overview of the stack to use - Obsidian + Claude + Cursor + InfraNodus graph / MCP server 23:15 Running gap analysis on LLM Wiki using the MCP server - without using the graph interface