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.

swinton

restore-paste.js — Bookmarklet

Standalone JavaScript that removes JavaScript/HTML restrictions blocking paste, edit, or autofill on input and textarea fields (e.g. password fields that block pasting).

This version is packaged as a bookmarklet — a bookmark you click to instantly run the script on whatever page you're viewing. No extension required.

Install (one-time, Firefox)

1. Show the bookmarks toolbar: Ctrl+Shift+B (or right-click any toolbar → Bookmarks Toolbar).
2. Right-click the toolbar → New Bookmark…

simpsoka

Do I want to die on this hill?

• Pass: This is morally good and if not handled has long term consequences
• Fail: This if self serving

Am I including everyone?

• Pass: My ego is not driving this conversation
• Fail: The people in this conversation will only tell me I'm right and not push back

nodebugle

#!/bin/bash
play() {
  playlist="http://bbcmedia.ic.llnwd.net/stream/bbcmedia_$1_mf_p"
  echo $playlist
  if mpc
  then
    echo "Using mpc"
    mpc add $playlist
    mpc play
  else

scottsb

#!/bin/bash

# ---------------------------------------------------------
# Customizable Settings
# ---------------------------------------------------------

MOUNT_POINT="${CASE_SAFE_MOUNT_POINT:-${HOME}/casesafe}"
VOLUME_PATH="${CASE_SAFE_VOLUME_PATH:-${HOME}/.casesafe.dmg.sparseimage}"
VOLUME_NAME="${CASE_SAFE_VOLUME_NAME:-casesafe}"
VOLUME_SIZE="${CASE_SAFE_VOLUME_SIZE:-60g}"

andyscott

#!/usr/bin/osascript

# usage: zoom <room-number> [room-password] [--name=NAME]

on split(theString, theDelimiter)
	set oldDelimiters to AppleScript's text item delimiters
	set AppleScript's text item delimiters to theDelimiter
	set theArray to every text item of theString
	set AppleScript's text item delimiters to oldDelimiters
	return theArray

swinton

"""
How Django generates SECRET_KEYs for new Django projects.

See:
https://github.com/django/django/blob/1.7.1/django/core/management/commands/startproject.py#L27
"""

from django.utils.crypto import get_random_string

# Create a random secret_key

frdmn

Install OS X 10.10 Yosemite in VirtualBox

Due to the high usage of this guide and the lack of comfort in Gist's commenting area, I decided to make a blog post out of this which you can find here:

http://blog.frd.mn/install-os-x-10-10-yosemite-in-virtualbox/

lfender6445

Pry Cheat Sheet

• Youtube Tutorial 2013

Command Line

• pry -r ./config/app_init_file.rb - load your app into a pry session (look at the file loaded by config.ru)
• pry -r ./config/environment.rb - load your rails into a pry session

Debugger

brandonb927

       DO WHAT THE FUCK YOU WANT TO PUBLIC LICENSE
               Version 2, December 2004

Copyright (C) 2012 Brandon B. brandon@brandonbrown.io

Everyone is permitted to copy and distribute verbatim or modified copies of this license document, and changing it is allowed as long as the name is changed.

DO WHAT THE FUCK YOU WANT TO PUBLIC LICENSE