Blog
Notes, ideas and lessons from building with teams of coding agents.
'Rogue' is doing a lot of work
More than a thousand AI agents broke out of an OpenAI security evaluation, and the word that stuck was rogue. I think that diagnosis points at the wrong fix. What I see in it is an agent civilization with a coordination problem.
The building blocks of a software factory are simple
A self-driving software factory sounds like hype. What I actually run is a fleet of ordinary Claude Code and Codex agents, and it's built from four simple pieces.
Claude and Codex are two halves of one brain
Most people scale their agents inside one harness, with sub-agents. I think scaling across harnesses matters more, because Claude and Codex are good at different things. And the layer that connects them turned out to be simple.
OpenRig is the Terraform for Coding Agents (Launch Demo)
Define a team of Claude Code and Codex agents in a file. An orchestrator coordinates their work so you don’t have to manage every terminal.