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.

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.

These are very disorganized because I wrote these as I reversed various bits and pieces. If I'm wrong about something, please let me know so I can forget to update this.

For me this was the Achilles' heel, Crackproof creates a debug log if a certain folder in %temp% is present. The folder name is 12 hex characters long, and different between executables. The easiest way to find it is to break on/hook CreateFileW.

Once you create it, any Crackproofed modules will spit out logs as they unpack. Lines contain status codes, indicating roughly what the unpacker is doing at a given time (see below). Lines with additional debug information are also included, sometimes they can be very handy.