karpathy

LLM Wiki

A pattern for building personal knowledge bases using LLMs.

This is an idea file, it is designed to be copy pasted to your own LLM Agent (e.g. OpenAI Codex, Claude Code, OpenCode / Pi, or etc.). Its goal is to communicate the high level idea, but your agent will build out the specifics in collaboration with you.

The core idea

Most people's experience with LLMs and documents looks like RAG: you upload a collection of files, the LLM retrieves relevant chunks at query time, and generates an answer. This works, but the LLM is rediscovering knowledge from scratch on every question. There's no accumulation. Ask a subtle question that requires synthesizing five documents, and the LLM has to find and piece together the relevant fragments every time. Nothing is built up. NotebookLM, ChatGPT file uploads, and most RAG systems work this way.

nickbudi

Cygwin Git + VS Code compatibility

Thanks and credit to mattn and ferreus on GitHub.

Also check out Developing on WSL and/or wslpath (Windows 10 Build 17046 or later) if you're using the Windows Subsystem for Linux.

dfontana

What's this?

Instructions to obtain ZSH on a windows environment, without the input funny business presented by some other attempted solutions.

The final result is ZSH running on a mintty terminal, emulated by cygwin, and being handled by the popular cmder.

Why is this here?

For the benefit of myself and others. I've already followed these instructions twice. It took me hours to figure all this out, maybe someone else can save a few.

What exactly is covered?

• Installing and setting up cmder