// all posts
5 posts
Onboarding-as-code: a local, free Obsidian architecture that gives AI agents persistent memory, semantic search, and self-onboarding — in 15 minutes.
An introduction to VOID.LOG — a space for security research, AI systems, and the patterns that connect them.
A survey of the attack surface introduced by large language models — from prompt injection to model theft, and the defenses that actually work.
Cutting through the noise around threat intel — what separates actionable intelligence from indicator dumps, and how to build a useful picture of adversary behavior.
Applying traditional reverse engineering thinking to neural network weights — what you can learn, what remains opaque, and what tools exist.