Ken-B

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.

this assumes you have installed Python 3 via Anaconda distribution.

Make Python 2 kernel available to Jupyter Notebook

open a terminal and create a new python 2 environment: conda create -n py27 python=2.7
activate the environment: linux source activate py27 or windows activate py27
install the kernel in the env: conda install notebook ipykernel
install the kernel for outside the env: ipython kernel install --user
close the env: source deactivate

Locate the section for your github remote in the .git/config file. It looks like this:

[remote "origin"]
	fetch = +refs/heads/*:refs/remotes/origin/*
	url = git@github.com:joyent/node.git

Now add the line fetch = +refs/pull/*/head:refs/remotes/origin/pr/* to this section. Obviously, change the github url to match your project's URL. It ends up looking like this:

These notes are based on Patrick O'Leary's video that he kindly created when I was wingeing about problems with git. It is best to watch the video first then go back through these notes to recall the steps.

Create a clone of the JuliaLang repository

git clone git://github.com/JuliaLang/julia.git

or

git clone git://github.com/JuliaLang/julia.git myjulia

	#
	# linter.py
	# Linter for SublimeLinter3, a code checking framework for Sublime Text 3
	#
	# Written by Tomas Krehlik
	# Copyright (c) 2014 Tomas Krehlik
	#
	# License: MIT
	#

	function graythresh_kmeans(img)
	# Note, without the type annotation, this is slow because
	# mean appears not to be type stable here
	thresh = mean(img)::Gray{Float32}

	while true
	lastthresh = thresh
	nplus = 0
	nminus = 0
	splus = zero(typeof(thresh))