Alexandre Girard Davila alx

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.

Keybase proof

I hereby claim:

I am alx on github.
I am alxgirard (https://keybase.io/alxgirard) on keybase.
I have a public key whose fingerprint is B670 AE8A 90AE 2345 2126 1686 DB21 3723 D0B0 8480

To claim this, I am signing this object:

Keybase proof

I hereby claim:

I am alx on github.
I am alxgirard (https://keybase.io/alxgirard) on keybase.
I have a public key whose fingerprint is C681 143F 0F3D 468D A4FC 8485 D9E6 EC82 B82D 73A1

To claim this, I am signing this object:

	{
	"name": "Youtube url post to linkedin",
	"nodes": [
	{
	"parameters": {
	"formTitle": "Youtube transcribe",
	"formFields": {
	"values": [
	{
	"fieldLabel": "Youtube URL",

	#!/bin/bash
	adduser webui
	apt update && apt install wget git python3 python3-venv ffmpeg libsm6 libxext6 -y
	su webui
	cd /home/webui/
	git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui.git
	cd stable-diffusion-webui/models/Stable-diffusion
	wget https://huggingface.co/CompVis/stable-diffusion-v-1-4-original/resolve/main/sd-v1-4.ckpt
	cd ../../
	sed -i "s/#export COMMANDLINE_ARGS=\"\"/export COMMANDLINE_ARGS=\"--listen --skip-torch-cuda-test --share\"/g" webui-user.sh

	import feedparser
	from datetime import datetime, timedelta, timezone
	from dateutil import parser

	import requests
	from bs4 import BeautifulSoup
	import re

	import urllib.parse

	// ==UserScript==
	// @name OpenAI GPT-2 Detector
	// @namespace https://www.jolibrain.com/demo/openai-gpt2-detector-userscript
	// @description Paragraph of text reports the GPT-2 log prob of that text
	// @author Alexandre Girard <alex.girard@jolibrain.com>
	// @version 1.2
	// @grant none
	// @include https://en.wikipedia.org/wiki/*
	// ==/UserScript==

	arxiv
	from requests import sessions
	from pprint import pprint
	from rocketchat_API.rocketchat import RocketChat
	import arxiv
	import re

	regex = r"^https://arxiv\.org\/.\/(\d+\.\d+(?:v1)?).$"

	login = ""

	var fs = require('fs');
	var moment = require('moment');
	var DD = require('deepdetect-js');

	const dd = new DD({
	host: 'localhost',
	port: 18104,
	path: '/api/private/predict'
	})

	/* The API controller
	Exports 3 methods:
	* post - Creates a new thread
	* list - Returns a list of threads
	* show - Displays a thread and its posts
	*/


	var Thread = require('../models/thread.js');
	var Post = require('../models/post.js');