Karpathy's Idea Realized: OpenKB Auto-Compiles Docs into a Wiki
dr_cintas · x · 2026-08-12
Based on an idea proposed by Andrej Karpathy—having an LLM continuously compile everything it reads into a growing wiki rather than starting from scratch per query—a developer has built the open-source project OpenKB.
Key features of OpenKB include:
- Structured Knowledge Accumulation: Compiles raw documents into summaries, concept pages, and entity pages that are cross-linked and kept in sync, allowing knowledge to accumulate instead of being rediscovered.
- Vectorless Retrieval: Powered by PageIndex's vectorless, reasoning-based retrieval technology for long documents.
- Multimodality & Standards: Native multimodal support and adherence to the Google Open Knowledge Format (OKF) specification.
Related event: OpenKB Realizes Karpathy's Idea of Growing LLM Wiki(2 posts)→
More from coding & agent
- Berkeley Talks on AI Agents: Harness and Skill Engineering Becoming a Science — seanwbren · 2026-08-12
- Build High-Fidelity Video Scenes for Under $2 Using Code Skeletons and Video Models — mattshumer_ · 2026-08-12
- A Minimalist AI Coding Workflow: Auditing Sentry Instrumentation via Codex — zeeg · 2026-08-12
- How I Orchestrate 19 AI Agents Across 86 Data Sources to Generate Sales Intelligence Reports — Active-Ad9071 · 2026-08-12
- Serverless Framework Introduces Stateless MCP Server Deployment — DavidWells · 2026-08-12
- Stop Hoarding Agents: Why a Structured 4-Agent Setup Beats 20 in Parallel — PrajwalTomar_ · 2026-08-12