brx b0r1sp

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.

OS X Screencast to animated GIF

This gist shows how to create a GIF screencast using only free OS X tools: QuickTime, ffmpeg, and gifsicle.

Instructions

To capture the video (filesize: 19MB), using the free "QuickTime Player" application:

	Give this to your LLM of choice.
	Don't like python? ask it to build it in any language you prefer.

	Note: I am running this on Linux, but same thing can be done for Windows or macOS, too.
	----

	Build a "render-html-mcp" — MCP Server for Rendering HTML in Desktop Popup Windows
	Overview
	Create a Python MCP server (using the mcp package's FastMCP framework) that pops up a standalone desktop browser window displaying HTML/markdown content and returns immediately. The window stays open until the user closes it. The browser process is fully detached (start_new_session=True, stdio to /dev/null) so it outlives the MCP call.
	Project Structure

	# Not extensively tested
	# Put this script in the action_plugins directory of your playbook directory
	# If you have issues, please report it in the comments (or fork and fix)

	# Usage:
	# - name: "Ask the user if we should continue."
	# action: ask_key prompt="Continue? Yes / No / Random (y/n/r)?" accepted_keys="['y', 'n', 'r']"
	# register: answer
	#
	# The pressed key is now in answer.key

	#!/usr/bin/env bash

	DOCKER=`which docker`

	usage()
	{
	echo "Usage: $(basename $0) [-l num] IMAGE"
	exit 0
	}

	#!/usr/bin/env bash
	set -o errexit
	set -o nounset
	set -o pipefail

	# Automatically update your CloudFlare DNS record to the IP, Dynamic DNS
	# Can retrieve cloudflare Domain id and list zone's, because, lazy

	# Place at:
	# /usr/local/bin/cf-ddns.sh