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.

POWERFULMOVES

Claude Agent SDK Technical Specification

Version: 1.0.0 Protocol Version: 2024-11-05 Last Updated: 2026-01-10

Table of Contents

1. Overview
2. Architecture

agokrani

'system': 
[
  {
     'type': 'text', 
     'text': "You are Claude Code, Anthropic's official CLI for Claude.", 
     'cache_control': {'type': 'ephemeral'}
  }, 
  {
     'type': 'text',
     'text': 'You are an interactive CLI tool that helps users with software engineering tasks. 

shime

What?

Most basic example of authenticating with Github in node.

How?

Clone this gist, change keys inside config.js and then hit npm install && node app.js.

Done?

mikeal

I've known people at nodejitsu for years, since before the company even existed. I still consider many of them friends. That said, somebody over there has lost their mind.

Trademarks are an important part of open source. They protect the integrity of the trust that is built by any project. A classic example of why this is the case is Firefox. Suppose that a malware producer takes the Firefox codebase, which is free and open source, packages up their malware with it and then releases it as "Firefox". Then they buy search advertising and suddenly their bad and malicious version of Firefox is the first result on search engines across the web. This is clearly a bad thing for Firefox and open source everywhere, but what can Mozilla do to protect their community of users?

They can't enforce a software license since the use is permitted under the Mozilla Public License. They can, however, enforce on these hypothetical bad actors using their trademark on the word "Fi