Skip to content
Building release 1

Supervise AI coding agents. 
Stay in the loop.

Self-hosted: agents work on your repository issues in isolated workspaces while you answer questions, intervene, review and approve every push from one dashboard on every device. The command-line tool is whr.

Read the design Set it up

A supervisor with a dashboard, not an IDE

Tasks, runs, workspaces and environments are separate objects you watch and steer from the dashboard on any device or the CLI. Attaching or detaching an editor never interrupts the agent.

Human in the loop

Agents raise decisions: questions, approvals and reviews. You answer them from your phone, the web app or the CLI.

Isolated by default

Apple Container on an Apple-silicon Mac mini comes first: each workspace’s environment, a lightweight VM that its agents share, out to the internet only through an allowlist proxy, with short-lived, per-run forge credentials.

Approval boundaries are policy

Agents commit inside their environment; the host never runs git there. Their commits leave as a bundle, are checked against the supervisor’s mirror and are pushed only after you approve the exact commit. Merge, tag, release and deploy stay with you.

One service layer

whr and the web UI share the same service layer, so anything you can click you can script.

Traceable work

Every commit can carry its issue, task, run and the AI model that helped, so history shows who or what did what.

What goes wrong when agents run unsupervised

Each failure has an answer in the design. Release 1 is still being built: these are the intended behaviour, not a track record.

  • Nothing stops a push. Here an agent’s commits go to the forge only after you approve the exact commit; merge, tag, release and deploy stay with you. See security.
  • No isolation. Here each workspace has its own environment, and its agents reach the internet only through an allowlist proxy. See the design.
  • Secrets are within reach. Here forge credentials are short-lived and per run, and whr never handles a subscription login. See vendor terms.
  • No way to follow or stop a run. Here runs and decisions show in the dashboard and the CLI, where you can steer or stop them. See daily use.
  • One login shared by many agents. Here the agents of one workspace share that workspace’s sign-in (you sign in once per environment), and only you start runs. Several sessions on one sign-in: open (D42, #82). See vendor terms.