Jeremiah Goerdt CaptainKraft

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 D3D11

Minimal D3D11 reference implementation: An uncluttered Direct3D 11 setup + basic rendering primer and API familiarizer. Complete, runnable Windows application contained in a single function and laid out in a linear, step-by-step fashion that should be easy to follow from the code alone. ~200 LOC. No modern C++ / OOP or (other) obscuring cruft. View on YouTube

Other gists in this series:

Minimal D3D11 part 2: instancing

	// example how to set up D3D11 rendering on Windows in C

	#define COBJMACROS
	#define WIN32_LEAN_AND_MEAN
	#include <windows.h>
	#include <d3d11.h>
	#include <dxgi1_3.h>
	#include <d3dcompiler.h>
	#include <dxgidebug.h>

	// --- Library --------------------------------------------------------------------------
	#include <string.h>
	#include <assert.h>

	struct ui_box {int x,y,w,h;};
	typedef unsigned long long ui_id;
	struct ui_node {
	int parent;
	int lst, end, nxt;
	int siz[2];