OpenKB Realizes Karpathy's Idea: Turning LLMs into a Growing Wiki
dr_cintas · x · 2026-08-12
Andrej Karpathy once mused about an LLM that compiles everything it reads into a growing wiki rather than starting from scratch per query. A developer has now built exactly that: OpenKB.
How OpenKB works:
- Knowledge Accumulation: When a file is dropped in, the LLM reads it, writes a summary, cross-references it with existing concepts, and updates the wiki.
- No Rediscovery: Subsequent queries pull from the compiled knowledge base rather than initiating a fresh search, preventing knowledge loss.
- Indexing: Instead of using a vector database, the project tree-indexes long PDFs rather than chunking and embedding them.
Related event: OpenKB Realizes Karpathy's Idea of Growing LLM Wiki(2 posts)→
More from coding & agent
- AI Agent Preferences Reshape Dev Ecosystem: Drizzle Surpasses Prisma in NPM Downloads — tristanbob · 2026-08-12
- Developer Shares Practice of Using GitHub Copilot for Multi-Model Code Reviews — DanWahlin · 2026-08-12
- Expert Warns AI Watermarks Could Increase Code Entropy, Harming Coding Agents — wightmanr · 2026-08-12
- Claude Code Skills Catalog Adds NTM Multi-Agent Orchestration Skills, Downloads Exceed 80K — doodlestein · 2026-08-12
- Local Agent Memory Bottleneck: Vector Search Fails, Structured Data Models Emerge — CopyPasteVeteran · 2026-08-12
- LTX-2.5 Video Model Released: Dev Deploys 10 Subagents Overnight on Mac — cocktailpeanut · 2026-08-12