Etienne de Bruin etdebruin

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.

Integration Testing with Capybara

Integration testing is awesome. Years ago, running integration tests was painful, slow, and they were so brittle that every change to the codebase broke the test suite.

Today it's a different story. We have amazing tools that make a tough job much easier. Let's check them out.

Background on Integration Testing

Integration tests are critically important because they exercise your application just like a real user. They therefore depend on the full stack from your models up through your controllers, helpers, view templates, web server, database, and middleware.

	#!/bin/bash

	# Initialize MySQL database.
	# ADD this file into the container via Dockerfile.
	# Assuming you specify a VOLUME ["/var/lib/mysql"] or `-v /var/lib/mysql` on the `docker run` command…
	# Once built, do e.g. `docker run your_image /path/to/docker-mysql-initialize.sh`
	# Again, make sure MySQL is persisting data outside the container for this to have any effect.

	set -e
	set -x

	require "sinatra/base"
	require "sinatra/namespace"
	require "multi_json"

	require "api/authentication"
	require "api/error_handling"
	require "api/pagination"

	module Api
	class Base < ::Sinatra::Base

	#########################################
	# Very basic rack application showing how to use a router based on the uri
	# and how to process requests based on the HTTP method used.
	#
	# Usage:
	# $ rackup rack_example.ru
	#
	# $ curl -X POST -d 'title=retire&body=I should retire in a nice island' localhost:9292/ideas
	# $ curl -X POST -d 'title=ask Satish for a gift&body=Find a way to get Satish to send me a gift' localhost:9292/ideas
	# $ curl localhost:9292/ideas