AI Labs — hands-on, not hand-wavy.
Self-contained, runnable exercises. Most ship for a laptop with a free local LLM (Ollama, llama.cpp); a few require a single rented GPU. No vendor sandboxes, no playgrounds — real tools, real workflows.
Labs in production
The lab catalogue is opening.
The lab infrastructure is live (CPT, admin UI, single-lab template, this archive). First labs are being written — prompt-injection sandbox, RAG with Ollama, MCP server build-and-break. Subscribe to the newsletter or follow the AI track for releases.
AI Security track →