Projects
Things I'm building.
An open protocol for how humans and AI agents should earn each other's trust, and smaller tools for the way attention and execution actually work.
- Recede Open protocol · TypeScript + Python · Apache-2.0
An open protocol for earned trust between humans and AI agents. Trust is a trajectory, not a checkpoint: agents earn autonomy from evidence, and human review recedes exactly where it's warranted, with a receipt for every action you no longer read. One protocol, many flows; it lands first in the everyday SDLC.
- Clutch Open source · built in public
An execution harness for ADHD, shipped as a Claude Code plugin. ADHD is a performance problem, not a knowledge one: the intent is there, it just does not reach the wheels at the moment you act. Clutch is the transmission. It catches a stalled task and hands back one small move, and as of v1.3 it engages on its own.
- Quest Board PWA · for my family
A gamified daily-routine board that turns my sons' mornings into something they can run themselves, externalizing the executive-function scaffolding that doesn't come for free.
- JavOS Personal system
A small operating system for my own attention: an agentic, AI-first workspace for the decisions, ideas, and follow-through that need somewhere to live outside my head.