Luis Cavalcante luisffc

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.

tmux shortcuts & cheatsheet

start new:

tmux

start new with session name:

tmux new -s myname

	#!/bin/bash

	sudo su
	yum update -y
	yum install yum-plugin-copr -y
	yum copr enable @caddy/caddy -y
	yum install caddy -y
	printf ":80\nrespond \"Hello from $HOSTNAME\"" > /etc/caddy/Caddyfile
	caddy start --config /etc/caddy/Caddyfile

	// 🔥 Node 7.6 has async/await! Here is a quick run down on how async/await works

	const axios = require('axios'); // promised based requests - like fetch()

	function getCoffee() {
	return new Promise(resolve => {
	setTimeout(() => resolve('☕'), 2000); // it takes 2 seconds to make coffee
	});
	}


	import java.security.Key;
	import java.util.Properties;

	import javax.crypto.Cipher;
	import javax.crypto.spec.SecretKeySpec;
	import javax.persistence.AttributeConverter;
	import javax.persistence.Converter;

	import org.slf4j.Logger;

	import select
	import datetime

	import psycopg2
	import psycopg2.extensions

	from sqlalchemy import create_engine, text

	engine = create_engine("postgresql+psycopg2://vagrant@/postgres")

	from ansible import playbook, callbacks

	# uncomment the following to enable silent running on the playbook call
	# this monkey-patches the display method on the callbacks module
	# callbacks.display = lambda a,*ka: None

	# the meat of the meal. run a playbook on a path with a hosts file and ssh key
	def run_playbook(playbook_path, hosts_path, key_file):
	stats = callbacks.AggregateStats()
	playbook_cb = callbacks.PlaybookCallbacks(verbose=0)