Pascal Andy pascalandy

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.

Minimal Prompt Chainables

Sequential prompt chaining in one method with context and output back-referencing.

Files

main.py - start here - full example using MinimalChainable from chain.py to build a sequential prompt chain
chain.py - contains zero library minimal prompt chain class
chain_test.py - tests for chain.py, you can ignore this
requirements.py - python requirements

Setup

Get k3s with OpenFaaS using k3sup and multipass

multipass from Canoncial is like Docker Desktop, but for Ubuntu and works on MacOS, Linux and Windows.

Use-case:

We can get a Kubernetes cluster with k3s in a very short period of time. We can use this for workshops and building labs, and for R&D, including testing.

Get multipass.run

How to setup an Outline VPN Server on Ubuntu 16.04 Server

This guide will show you how to install Outline Server on an Ubuntu 16.04 Server, use Outline Manager for Windows and connect to your Outline Server on Windows and Anroid.

Install Outline Manager

Outline Manager supports Windows, macOS and Linux.

Outline Manager for Windows

Docker Cli Tips and Tricks

My favorite command line options and best practices

Bash/zsh completion
use Tab to populate command line, narrows down options as you go, fills in container names, ID's, images, volumes. Huge time saver.

3 x Cluster on AWS

manager1
worker1
worker2

Docker-machine commands

Use docker-machine to lay down 3 x hosts on AWS using docker-engine 1.12

docker-machine create --driver amazonec2  --amazonec2-private-address-only --engine-install-url https://test.docker.com/ levenson-docker-swarm-manager1 
docker-machine create --driver amazonec2  --amazonec2-private-address-only --engine-install-url https://test.docker.com/ levenson-docker-swarm-worker1

Full blog post can be found here: http://pnommensen.com/2014/09/07/high-performance-ghost-configuration-with-nginx/

Ghost is an open source platform for blogging founded by John O'Nolan and Hannah Wolfe. It's a node.js application and therefore works great in conjunction with nginx. This guide will will help you create a high performance nginx virtual host configuration for Ghost.

"Don't use #nodejs for static content" - @trevnorris. If #nginx isn't sitting in front of your node server, you're probably doing it wrong.
— Bryan Hughes (@nebrius) August 30, 2014

The node.js application runs on a port on your server

	# NOT FOR SHELL SCRIPT, but rather just for quick copy paste
	# this is a copy-paste version with defaults of the full shell script docker-xenial.sh which is below this one in gist.

	apt-get -y install apt-transport-https ca-certificates curl && \
	curl -fsSL https://download.docker.com/linux/ubuntu/gpg \| sudo apt-key add - && \
	add-apt-repository "deb [arch=amd64] https://download.docker.com/linux/ubuntu $(lsb_release -cs) stable" && \
	service lxcfs stop && apt-get remove -y -q lxc-common lxcfs lxd lxd-client && \
	apt-get update -q && \
	apt-get install -y -q docker-ce && \
	printf '{ "userns-remap" : "default" , "storage-driver" : "overlay2" }' > /etc/docker/daemon.json && \

	# sets the proxy cache path location, max size 2g
	proxy_cache_path /data/nginx/cache levels=1:2 keys_zone=STATIC:100m inactive=24h max_size=2g;

	# transfers the `Host` header to the backend
	proxy_set_header Host $host;

	# uses the defined STATIC cache zone
	proxy_cache STATIC;

	# cache 200 10 minutes, 404 1 minute, others status codes not cached

	module.exports = {
	plugins: [require("tailwindcss"), require("autoprefixer")],
	}