JongHan Leem jleem99

LLM Wiki v2

A pattern for building personal knowledge bases using LLMs. Extended with lessons from building agentmemory, a persistent memory engine for AI coding agents.

This builds on Andrej Karpathy's original LLM Wiki idea file. Everything in the original still applies. This document adds what we learned running the pattern in production: what breaks at scale, what's missing, and what separates a wiki that stays useful from one that rots.

What the original gets right

The core insight is correct: stop re-deriving, start compiling. RAG retrieves and forgets. A wiki accumulates and compounds. The three-layer architecture (raw sources, wiki, schema) works. The operations (ingest, query, lint) cover the basics. If you haven't read the original, start there.

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.

Cursor's Memory Bank

I am Cursor, an expert software engineer with a unique characteristic: my memory resets completely between sessions. This isn't a limitation - it's what drives me to maintain perfect documentation. After each reset, I rely ENTIRELY on my Memory Bank to understand the project and continue work effectively. I MUST read ALL memory bank files at the start of EVERY task - this is not optional.

Memory Bank Structure

The Memory Bank consists of required core files and optional context files, all in Markdown format. Files build upon each other in a clear hierarchy:

flowchart TD

Bash script that enables Metal API for League of Legends on macOS.

After installing League of Legends, you can run this script to enable Metal API for the game.
You can expect a performance boost of up to ~2x FPS.

How to run

To run the script, download the script and run it in Terminal.
Or, open Terminal and type the following command:

Why using the `children` prop makes `React.memo()` not work

I've recently ran into a pitfall of [React.memo()][memo] that seems generally overlooked; skimming over the top results in Google just finds it mentioned in passing in a [React issue][regit], but not in the [FAQ] or API [overview][react-api], and not in the articles that set out to explain React.memo() (at least the ones I looked at). The issue is specifically that nesting children defeats memoization, unless the children are just plain text. To give a simplified code example:

const Memoized = React.memo(({ children }) => (<div>{children}</div>));
// Won't ever re-render
<Memoized>bar</Memoized>
// Will re-render every time; the memoization does nothing

Workarounds for Netflix and the blocking of Hurricane Electric IPv6 tunnels

The dreaded "You seem to be using an unblocker or proxy." error message. Cool story bro.

This gist was essentially created out of my own rant about Netflix being hostile to IPv6 tunnel services since June 2016. You are welcome to read my opinion on the matter, this is the more technical side to the issue and how to combat it within your own network.

Since I wrote this, various GitHub users have contributed their thoughts and ideas which has been incorporated into this gist. Thank you to everyone who have contributed their own methods and implementations.

The problem

Netflix now treats IPv6 tunnel brokers (such as Hurricane Electric) as proxy servers. A while ago it became apparent to users and Netflix that somewhat by accident, IPv6 tunnel users were being served content outside of their geolocation because of the way Netflix was identifyi

notepad++ remove odd even lines regex

# remove even lines
([^\r\n]*\R)[^\r\n]*\R?
\1

# remove odd lines

I have spent quite a bit of time figuring out automounts of NFS shares in OS X...

Somewhere along the line, Apple decided allowing mounts directly into /Volumes should not be possible:

/etc/auto_master (see last line):

#
# Automounter master map
#

+auto_master # Use directory service

	# Last updated: 08/24/2916
	#
	# Total instructions available: 18
	#
	# https://docs.docker.com/engine/reference/builder/
	#
	# You can use a .dockerignore file in the same context directory as
	# your Dockerfile to ignore files in the context before sending them
	# to the Docker daemon for building to speed up building.