Skip to main content

Introduction

Note

The goal is not to publish finished theory. It is to keep useful technical knowledge reproducible, current and honest about what is known, what is inferred and what still needs testing.

Start with what you need​

  • Malware analysis: file triage, PE imports and hash reputation checks. Detection engineering turns those observations into rules.
  • Local LLMs: hardware, llama.cpp, multi-GPU inference, context sizing, speculative decoding, benchmarking and agent clients.
  • Prompts and learning: copyable prompts for technical work, study and teaching.
  • Runbooks: step-by-step procedures for tasks where the order matters.
  • Homelab: the isolated network and infrastructure behind the systems in these notes.
  • Focus and ADHD: how I plan, start and restart work.

How the site is organised​

  • Docs are maintained notes. They change when testing or new evidence changes what I know, and each one shows when it was last reviewed.
  • Journal entries are dated and stay as written, with a correction note when something turns out wrong.
  • Runbooks are procedures from a known starting state to a result you can check. Each step ends with a check you can see.
  • Topics groups everything by subject, and is the quickest way in.

How the notes are written​

Evidence and interpretation stay apart. Uncertainty stays visible, and claims should be reproducible where possible. When a result depends on hardware, software version, configuration or test conditions, that context stays with the result.

Read How these notes are written for the full conventions.

Pick a starting point​