HN user

mutant

332 karma
Posts14
Comments399
View on HN

you built a website using another human's name, and then made a terrible wiki that doesnt even cover the subject matter well

i think this should be the most important subject on the site

https://andrej-karpathy.com/#/concepts/llm-wiki.md

and im pretty sure my partner knows more about llmwikis listening to me babble about them more than she would learn on these pages

the top nav is abysmal the cognee graph is useleess

''' Extracted Claims A useful LLM knowledge base can be maintained as a persistent markdown wiki. Raw sources, generated wiki pages, and schema/instructions should be separate layers. Ingest, query, and lint are the core operations. Query outputs can be filed back into the wiki when they contain reusable synthesis. Index and log files help both humans and agents navigate the evolving wiki. ''' this sis the most cogent thought on the subject and its on a stub

https://andrej-karpathy.com/#/source-notes/llm-wiki-gist.md

wow. termerity has a website

Yeah, I'm well aware of the pattern, and I'm aware of the Cloudflare code paper and I know that the overhead of MCP is still unneeded because you still have to have those open HTTP connections or local connections or whatever the case is. Those are completely unnecessary when your harness can use a tool like a CLI. It's way more efficient. You can explain it in less tokens or you can type script out the entire primitive so that you don't need the overhead.

I still don't see it. I appreciate that you do and that you made a product to deal with it but I just think it's still a bad pattern.