In development

Agent harness

A harness for running AI agents — reliably, and at scale.

The scaffolding around autonomous agents: orchestration, tools, isolation, and guardrails so agents do real work without falling over. Built with the same bar as hushwarren — robust by default, quiet when it's working.

Coming soon Get in touch →

// what it does

Orchestration

For multi-step, multi-agent workflows.

Safe execution

Sandboxed, observable, recoverable.

Composable tooling

Tooling that composes instead of fighting you.

Robust by default. Quiet when it's working.

In active development. Drop us a line if you'd like to follow along.

Get in touch →