Eunkwang Shin gracefullight

LLM Wiki v2

A pattern for building personal knowledge bases using LLMs. Extended with lessons from building agentmemory 20K+ Stars ⭐️, 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.

Getting Started with ChatGPT Codex + Mastra Agents

Step-by-Step Instructions

Open the ChatGPT Codex task setup panel. This is where you configure your environment before starting a task.
Locate the "Setup Script" field. You’ll see a note that internet access is disabled after the script runs.

FAQ on the xz-utils backdoor (CVE-2024-3094)

This is a living document. Everything in this document is made in good faith of being accurate, but like I just said; we don't yet know everything about what's going on.

Update: I've disabled comments as of 2025-01-26 to avoid everyone having notifications for something a year on if someone wants to suggest a correction. Folks are free to email to suggest corrections still, of course.

Background

macOS Internals

Understand your Mac and iPhone more deeply by tracing the evolution of Mac OS X from prelease to Swift. John Siracusa delivers the details.

Starting Points

How to use this gist

You've got two main options:

AsyncLocalStorage

어떤 문제가 있나요

우리의 서버로 하나의 request가 들어오면 서버는 응답을 처리하기 위해 필요한 여러 API를 호출하고 데이터를 가공하여 필요한 응답을 돌려줍니다. requset는 동시에 여러개 들어오기 때문에 우리는 request마다 어떤 작업을 수행했는지 그룹핑하여 모니터링하는 것이 필요합니다. 가장 일반적인 방법으로는 요청마다 requestID 를 생성해 할당하고 수행하는 작업마다 requestID 를 함께 기록하여 requestID 단위로 그룹핑해 확인하는 것입니다.

이러한 패턴은 Ruby on rails 나 Python Django 같은 플랫폼에서 매우 간단하게 구현할 수 있는데, 각 request는 하나의 thread에 할당되어 동기로 처리되기 때문에 requestID 와 같은 요청 정보는 thread context 에 저장하여 필요할 때 꺼내 사용하면 되기 때문입니다.

In the olden days, HTML was prepared by the server, and JavaScript was little more than a garnish, considered by some to have a soapy taste.

After a fashion, it was decided that sometimes our HTML is best rendered by JavaScript, running in a user's browser. While some would decry this new-found intimacy, the age of interactivity had begun.

But all was not right in the world. Somewhere along the way, we had slipped. Our pages went uncrawled by Bing, time to first meaningful paint grew faster than npm, and it became clear: something must be done.

And so it was decided that the applications first forged for the browser would also run on the server. We would render our HTML using the same logic on the server and the browser, and reap the advantages of both worlds. In a confusing series of events a name for this approach was agreed upon: Server-side rendering. What could go wrong?

In dark rooms, in hushed tones, we speak of colours.

	"""
	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

	"mcpServers": {
	"codex_subagent": {
	"command": "npx",
	"args": [
	"-y",
	"@openai/codex",
	"mcp",
	"-c model=\"gpt-5\"",
	"-c model_reasoning_effort=\"high\""
	]

	// I'm tired of extensions that automatically:
	// - show welcome pages / walkthroughs
	// - show release notes
	// - send telemetry
	// - recommend things
	//
	// This disables all of that stuff.
	// If you have more config, leave a comment so I can add it!!

	{