slaiyer

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.

jj Cheat Sheet for Git Users

🗣️ this is a useful guide to help you memorize + better understand how Jujutsu works, but it's not a replacement for learning the mental model or deeper strategies!

Core Concepts

No staging area — all changes in the working copy are automatically part of the current “change”
Changes vs commits — jj tracks “changes” (mutable) that become “commits” (immutable) when pushed
@ — a symbol pointing to the current working copy change (like HEAD, but for changes)
@- — the parent of @

Electron Apps Causing System-Wide Lag on Tahoe

This script detects apps with not yet updated versions of Electron.

Repo: https://github.com/tkafka/detect-electron-apps-on-mac

See:

PostgreSQL is Enough

See the new site: https://postgresisenough.dev

Learning Rust

Getting Started + Installation | Cheat Sheet

	# https://www.aloxaf.com/2024/02/manage_zsh_shell_with_atuin/

	export ATUIN_TMUX_POPUP=false
	# shellcheck disable=SC2034,SC2153,SC2086,SC2155

	# Above line is because shellcheck doesn't support zsh, per
	# https://github.com/koalaman/shellcheck/wiki/SC1071, and the ignore: param in
	# ludeeus/action-shellcheck only supports _directories_, not _files_. So
	# instead, we manually add any error the shellcheck step finds in the file to
	# the above line ...

	# MacOS Version 1.2
	# 08/14/2025
	# Requires MPV v0.38.0 or newer

	###############################################################################################################################################

	# Player GUI Options

	###############################################################################################################################################

	import contextlib
	import io
	import multiprocessing
	import traceback
	from multiprocessing.managers import SyncManager

	# For demo only
	import random
	import time

	{
	"title": "Quake-style Hotkey Window for Wezterm",
	"rules": [
	{
	"description": "Launch/show Wezterm if it is not in foreground",
	"manipulators": [
	{
	"type": "basic",
	"from": {
	"key_code": "escape",

	import datadog.trace.api.DDTags
	import io.opentracing.Span
	import io.opentracing.Tracer
	import io.opentracing.log.Fields
	import io.opentracing.tag.Tags
	import io.opentracing.util.GlobalTracer

	suspend fun <T : Any, O> T.coRunTraced(
	methodName: String,
	block: suspend T.(Span) -> O,