safessh
Open sourcePolicy-gated remote access that gives humans and AI agents the same named projects, approvals, and audit trail.
Selected systems
Project Showcase
Filter by technology
Aiva
Open sourceA personal assistant console built for the Motorola Razr cover display, backed by tools, memory, scheduling, and explicit action approval.
Mind Queue for Pi
Open sourceA durable project-wide queue that keeps useful thoughts available across coding-agent sessions without injecting them into every prompt.
Essence
Open sourceA portable, Markdown-first life-management system operated through coding-agent commands instead of another proprietary application.
Field notes & case studies
From the Writing Desk
The first field notes and case studies are on the way.
Receipts included
Engineering Decisions
Policy gates before remote execution
ACCEPTEDAgents get a constrained interface; operators keep credentials and approval.
Durable files over hidden application state
ACCEPTEDPortable data remains inspectable after the surrounding tool changes.
Unrestricted agent access to infrastructure
REJECTEDConvenience is not worth an invisible production trust boundary.
The engineer
About & Now
Background
I build AI systems for teams that need to trust what those systems do.
Years of shipping products end to end across mobile, web, and backend. Now a CTO and open-source builder of tools for working with agents.
Currently
Building: safer interfaces between agents and infrastructureExploring: useful memory without accumulated noiseWriting: practical notes from production-minded AI work
Interested in agent systems, architecture, or practical AI delivery?
Start a conversation