System

I run my work through a system of agents, rules and pipelines on a Mac Studio in my living room. This page is its public partition. Structure public, contents private: code from active projects ships only as generalized teaching artifacts, never raw.

What the components do

• claude-os. A distilled operating manual for any AI agent that works with me, plus ten skills earned from real failures: verifying done before claiming done, propagating corrections across a whole project, syncing three copies of every repo, running overnight builds to completion, writing in my voice, and more. Every rule exists because something failed twice.
• second-brain. A knowledge system on the LLM-wiki pattern: agents read raw notes and compile an interlinked wiki, so knowledge accumulates instead of being re-derived. Sensitive folders never leave the machine; a local model is the janitor at the door.
• finance-pipeline. Personal finance automation that runs entirely on my own machine. A local model does the categorising, and the documents are OCR'd on-device, so nothing gets uploaded anywhere.

Extracted artifacts

Coming here as tutorials with sample data: the parametric token methodology from the design-system case, a template agent-instructions file, and the setup script that builds a clean dev home. Each one generalized, marked SAMPLE where data would be.