David Higginbotham dhigginbotham

LLM Wiki v2

A pattern for building personal knowledge bases using LLMs. Extended with lessons from building agentmemory 10K 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.

Currently, Working on AKBP: Agent Knowledge Base Protocol based on my findings, a protocol for creating, updating, retrieving, and sharing durable knowledge across AI agents.

What the original gets right

1.) Download a Nerd Font

2.) Unzip and copy to ~/.fonts

3.) Run the command fc-cache -fv to manually rebuild the font cache

Docker Host Server Setup, w/ Basic Monitoring Tools

# Updates
apt-get update && apt-get install -y vim-nox git-core ufw curl atop htop build-essential libssl-dev linux-image-amd64 linux-headers-amd64

# Updates Profile init scripts
cd ~/
curl -sSL https://gist.githubusercontent.com/justsml/b667f158731fd054cd38/raw/5778dbb5d3d138ccf99ae1bf973457ce89661362/.bash_aliases > .bash_aliases_new
cat .bash_aliases_new >> .bash_aliases

	{
	"Export default": {
	"scope": "javascript,typescript,javascriptreact,typescriptreact",
	"prefix": "eid",
	"body": [
	"export { default } from './${TM_DIRECTORY/.[\\/](.)$$/$1/}'",
	"$2"
	],
	"description": "Import and export default in a single line"
	},

	const puppeteer = require('puppeteer');
	const proxyChain = require('proxy-chain');

	(async() => {
	const oldProxyUrl = 'http://bob:password123@proxy.example.com:8000';
	const newProxyUrl = await proxyChain.anonymizeProxy(oldProxyUrl);

	// Prints something like "http://127.0.0.1:45678"
	console.log(newProxyUrl);

	-- Create the table
	CREATE TABLE IF NOT EXISTS `StateBoundaries` (
	`State` varchar(10) DEFAULT NULL,
	`Name` varchar(255) DEFAULT NULL,
	`MinLat` varchar(50) DEFAULT NULL,
	`MaxLat` varchar(50) DEFAULT NULL,
	`MinLon` varchar(50) DEFAULT NULL,
	`MaxLon` varchar(50) DEFAULT NULL
	) ENGINE=InnoDB DEFAULT CHARSET=utf8;

	#!/bin/bash
	db=<db>
	collection_list="<collection1> <collection2> <collection3>"
	host=127.0.0.1
	port=<port>

	out_prefix=/Temp
	for collection in $collection_list; do
	echo $collection
	out_dir="${out_prefix}/${db}_${collection}/"

	#!/bin/sh -e
	#
	# NOTE: Since Docker 1.10 (February 4, 2016), it has been possible to configure the
	# Docker daemon using a JSON config file. On Linux, this file is normally located at
	# /etc/docker/daemon.json. You should use this JSON config method if you are running
	# a version of Docker that is at least 1.10!
	# Here is an example configuration that sets the docker0 bridge IP to 192.168.254.1/24:
	# {
	# "bip": "192.168.254.1/24"
	# }

	<html>
	<head>
	<title>JsTestBed</title>
	</head>
	<body>
	<h2>"Async" test</h2>
	<p>
	In JavaScript, asynchronous calls are a lie!<br />
	What actually happens is the setTimeout(func, 0); gets put on the end of the call list.<br />
	Any direct javascript functions will get prioritised and run before the function in the timeout.