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.

dabit3

See more of my writing here. Also check out Devin

In this post, I'll start from scratch and build up to OpenClaw's architecture step by step, showing how you could have invented it yourself from first principles, using nothing but a messaging API, an LLM, and the desire to make AI actually useful outside the chat window.

End goal: understand how persistent AI assistants work, so you can build your own (or become an OpenClaw power user).

First, let's establish the problem

When you use ChatGPT or Claude in a browser, there are several limitations:

dhh

# Create a new worktree and branch from within current git directory.
ga() {
  if [[ -z "$1" ]]; then
    echo "Usage: ga [branch name]"
    exit 1
  fi

  local branch="$1"
  local base="$(basename "$PWD")"
  local path="../${base}--${branch}"

perfectbase

/* eslint-disable @typescript-eslint/no-explicit-any */
import { type TRPCQueryOptions } from '@trpc/tanstack-react-query';
import { unstable_noStore } from 'next/cache';
import { Fragment, Suspense, type ReactNode } from 'react';
import { ErrorBoundary } from 'react-error-boundary';
import { HydrateClient, prefetch as prefetchTRPC } from '@/trpc/server';

type AwaitProps<T> =
  | {
      promise: Promise<T>;

devinrhode2

const { safeEnv } = require('./src/env')

// Note: NextConfig type allows any keys, so we are plucking the specific keys we want to type-check.
/**
 * @type {Pick<
 *   import('next').NextConfig,
 *   | 'webpack'
 *   | 'env'
 * >}
 */

AngusFu

// 粘贴到 chrome devtool console 中执行
// 使用场景 
//    1. 想复制只读文档的内容
//    2. 担心复制内容的行为被公司监听了

function removeAllEvents(event) {
  function removeEvent(el, type) {
    const listeners = (getEventListeners(el)[type] || []);

    el[`on${type}`.toLowerCase()] = null;

tannerlinsley

import { BroadcastChannel, createLeaderElection } from 'broadcast-channel'
import React from 'react'

const channels = {}

export function useBroadcastLeader(id = 'default') {
  const [isBroadcastLeader, setIsBroadcastLeader] = React.useState(false)

  React.useEffect(() => {
    if (!channels[id]) {

jamiebuilds

assert() (sometimes called invariant())

Instead of checks like:

if (value === null) {
  throw new Error("missing value")
}
doSomethingThatNeedsValue(value)

tannerlinsley

Replacing Create React App with the Next.js CLI

How dare you make a jab at Create React App!?

Firstly, Create React App is good. But it's a very rigid CLI, primarily designed for projects that require very little to no configuration. This makes it great for beginners and simple projects but unfortunately, this means that it's pretty non-extensible. Despite the involvement from big names and a ton of great devs, it has left me wanting a much better developer experience with a lot more polish when it comes to hot reloading, babel configuration, webpack configuration, etc. It's definitely simple and good, but not amazing.

Now, compare that experience to Next.js which for starters has a much larger team behind it provided by a world-class company (Vercel) who are all financially dedicated to making it the best DX you could imagine to build any React application. Next.js is the 💣-diggity. It has amazing docs, great support, can grow with your requirements into SSR or static site generation, etc.

So why

DeruiDENG

Using JSDOC-Based TypeScript

Get Started

Choose your editor

• WebStorm, Rider
  • Partial support, not enough intelli hints
  • Toggle on TypeScript language service
• VSCode