AI news story
Show HN: A Karpathy-style LLM wiki your agents maintain (Markdown and Git)
I shipped a wiki layer for AI agents that uses markdown + git as the source of truth, with a bleve (BM25) + SQLite index on top. No vector or graph db yet.It runs locally in ~/.wuphf/wiki/ and you can git clone it out if you want to take your knowled
Editor's take
A developer has created a local wiki system designed to serve as a persistent knowledge base for AI agents, leveraging Markdown files and Git for version control, with a BM25 index for retrieval.
This initiative addresses a critical challenge in agent development: maintaining context and learned information across sessions without relying on complex databases. By using familiar tools like Git, it lowers the barrier to entry for agents to build and access a structured knowledge graph, potentially enabling more sophisticated agent behavior beyond single-turn interactions and reducing the need for constant retraining of models like Llama 3 or Mistral.
Future developments will likely focus on integrating vector search for semantic retrieval and exploring how this local knowledge base can be shared or synchronized between multiple agents or even human users. The extent to which this system can scale and adapt to diverse agent tasks will determine its practical impact.
Signal score: 3
This event was corroborated by 6 independent sources. The signal score weighs cross-source corroboration, recency, source weight and topic salience. How we rank stories.
Original reporting
This story summarises reporting published by Hacker News. Read the original article at Hacker News.