Daiki Noda sys9kdr

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.

	//FindChangeByList.jsx
	//An InDesign CS6 JavaScript
	/*
	@@@BUILDINFO@@@ "FindChangeByList.jsx" 3.0.0 15 December 2009
	Mod by senki on 2015-06-29
	- added dialog with list of files under FindChangeSupport
	- and choose between. default is the first
	*/
	//Loads a series of tab-delimited strings from a text file, then performs a series
	//of find/change operations based on the strings read from the file.

	import subprocess

	get_line_by_line_texlive_dependencies = subprocess.run(
	[
	"apt-cache",
	"depends",
	"texlive-full"
	],
	universal_newlines=True,
	stdout=subprocess.PIPE

	<!DOCTYPE html>
	<html>
	<head>
	<meta charset="utf-8">
	<title>CSS Fizz Buzz</title>
	<style>
	#main li:nth-child(3n),
	#main li:nth-child(5n){
	font-size: 0;
	}

	@namespace url(http://www.w3.org/1999/xhtml);
	@-moz-document url-prefix(data:image/),
	regexp("(?:https?\|chrome)://.\.(?:jpe?g\|png\|gif\|bmp\|svg\|webp\|JPE?G\|PNG\|GIF\|BMP\|SVG\|WEBP)([?#].)?$") {
	body {
	background-image: url("data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAACAAAAAgCAAAAABWESUoAAAAHUlEQVQ4y2P4gQYeoAGGUQUjSgG6ALqGUQUjSgEAdjWwLh+tpFgAAAAASUVORK5CYII=") !important;
	}

	body > img[src][alt] {
	box-shadow: 0px 0px 2px 1px navy;
	}