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.

squarism

In the below keyboard shortcuts, I use the capital letters for reading clarity but this does not imply shift, if shift is needed, I will say shift. So ⌘ + D does not mean hold shift. ⌘ + Shift + D does of course.

Tabs and Windows

Function	Shortcut
New Tab	⌘ + T
Close Tab or Window	⌘ + W (same as many mac apps)
Go to Tab	⌘ + Number Key (ie: ⌘2 is 2nd tab)
Go to Split Pane by Direction	⌘ + Option + Arrow Key

olivierlacan

git config --global alias.hist "log --pretty=format:'%h %ad | %s%d [%an]' --graph --date=short"
git config --global alias.lol "log --graph --decorate --pretty=oneline --abbrev-commit --all"
git config --global alias.mylog "log --pretty=format:'%h %s [%an]' --graph"

To check that they've been added correctly, first run git config --list. You should see something like this in the midst of all your other configuration:

alias.hist=log --pretty=format:"%h %ad | %s%d [%an]" --graph --date=short