Romain Guerin pomeh

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.

The purpose of this document is to make recommendations on how to browse in a privacy and security conscious manner. This information is compiled from a number of sources, which are referenced throughout the document, as well as my own experiences with the described technologies.

I welcome contributions and comments on the information contained. Please see the How to Contribute section for information on contributing your own knowledge.

nodeconf slides - 2011

a list of slides from nodeconf

you may want to take a look at the jsconf-gist too!

JS Conf slides - 2011

Track A

Bytes and Blobs by David Flanagan
Conference Wifi Redux by Malte Ubi
Sashimi - https://github.com/cramforce/Sashimi
Run Your JS everywhere with Jellyfish by Adam Christian - http://jelly.io Project
Fighting Crime and Kicking Apps with Batman.js by Nick Small
Hello Jo by Dave Balmer - Project - http://joapp.com

	[
	{
	"keys": ["ctrl+w"],
	"command": "run_multiple",
	"args": {
	"commands": [
	{"command": "find_under_expand", "args": null, "context": "window"},
	{"command": "show_panel", "args": {"panel": "find"}, "context": "window"}
	]
	}

	#!/usr/bin/env ruby -w

	# pnginator.rb: pack a .js file into a PNG image with an HTML payload;
	# when saved with an .html extension and opened in a browser, the HTML extracts and executes
	# the javascript.

	# Usage: ruby pnginator.rb input.js output.png.html

	# By Gasman <http://matt.west.co.tt/>
	# from an original idea by Daeken: http://daeken.com/superpacking-js-demos

	//
	// Usage: require('./pid')("myapp");
	//
	var fs = require('fs');

	module.exports = function(appname){
	process.title = appname;

	var PID_FILE = "/usr/local/var/run/"+process.title+".pid";

	#!/bin/bash

	# Configure your account.
	USER=espadrine

	# Ask for the password.
	echo -n password\ ; read -s PASS

	curl -u $USER:$PASS -d status="$1" http://twitter.com/statuses/update.xml

	# Here a few bash one-liners that helped me analyze / fight a weak DOS attack against debuggable.com. Mostly for future reference.
	# The attacker was opening lots of tcp connections without sending data, I believe it's called a SYN flood, see: http://tools.ietf.org/html/rfc4987#section-3.2

	# Step 0: Check what is going on at port 80
	$ netstat -tan \| grep ':80 ' \| awk '{print $6}' \| sort \| uniq -c

	# Step 1: Increase the number of available fds
	$ ulimit -n 32000

	# Step 2: Restart your webserver, for me: