Nick Powell nmpowell

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.

Meta Prompting

In the Generative AI Age your ability to generate prompts is your ability to generate results.

Guide

Claude 3.5 Sonnet and o1 series models are recommended for meta prompting.

Replace {{user-input}} with your own input to generate prompts.

Use mp_*.txt as example user-inputs to see how to generate high quality prompts.

kamal

This documentation adds important additions to the docs for kamal deploy tool (see github.com/basecamp/kamal)

Destination flag

You can use kamal deploy --destination staging

This will read config/deploy.yml and config/deploy.staging.yml files, and also will read .env.staging file if it exists.

How to use:

./wordle.sh

Or try the unlimit mode:

The included script 'widevine-flash_armhf.sh' fetches a ChromeOS image for ARM and extracts the Widevine and Flash binaries, saving them in a compressed archive. Since it downloads a fairly large file (2Gb+ on disk after download) it is recommended that you run the script on a raspberry that has plenty of disk space.

The files in the compressed archive are copied to the folder /usr/lib/chromium-browser/ To run the file just type the following:

sudo ./widevine-flash_armhf.sh

Check out this tutorial: https://lemariva.com/blog/2020/06/raspberry-pi-amazon-prime-netflix-and-drm-solution

	# /// script
	# dependencies = [
	# "SpeechRecognition",
	# "mlx-whisper",
	# "pyaudio",
	# ]
	# ///

	import speech_recognition as sr
	import numpy as np

	<!DOCTYPE html>
	<html lang="en">
	<head>
	<meta charset="UTF-8">
	<meta name="viewport" content="width=device-width, initial-scale=1.0">
	<title>Fullscreen Image Viewer</title>
	<style>
	body {
	font-family: Arial, sans-serif;
	}

	const defaultTheme = require('tailwindcss/defaultTheme');

	module.exports = {
	purge: [
	'./vendor/laravel/framework/src/Illuminate/Pagination/resources/views/*.blade.php',
	'./storage/framework/views/*.php',
	'./resources/views/*/.blade.php',
	'./resources/js/*/.vue',
	],

	import argparse
	import importlib
	import logging
	import os
	import sys

	def main(args=None):
	scriptname = os.path.basename(__file__)
	parser = argparse.ArgumentParser(scriptname)
	levels = ('DEBUG', 'INFO', 'WARNING', 'ERROR', 'CRITICAL')

	#!/bin/bash

	# Bash script to control a TPLink HS110 or HS100 smart plug.

	# Usage: $0 -u username -p password -d "?" => list devices, and should output token.
	# Usage: $0 -t <token> -d <device_id> -o off => turn off..
	# Usage: $0 -t <token> -d <device_id> -o off => turn on..

	# Default action is to turn the first device found off.