Skip to main content

Parked runs wake sooner, plus a Requester filter for roadmap feedback

· 2 min read

Parked runs resume sooner when computer capacity frees up, roadmap feedback gains a Requester filter, and several routing and session issues are fixed.

Runs waiting on computer capacity now resume much sooner, and freed computers go straight to the next run in line. This update also adds a Requester filter to roadmap feedback, a cleaner ship maintenance settings page, and a batch of routing and session fixes.

Waiting for computer capacity​

  • Parked runs now wake automatically when computer capacity improves, a device comes online, or a computer updates. They also re-check periodically, so they resume sooner when quotas reset early.
  • Freed computers are offered to the first compatible run waiting in the queue, so machines no longer sit idle.
  • A parked run can now trigger a usage refresh on teammates' contributed computers, so shared capacity wakes up after quota resets or account updates.
  • Temporary computer-level refusals now expire, so routing waits and retries instead of failing straight away.

Routing and reliability​

  • Queued runs no longer get stuck. Older parked runs are re-checked when capacity is available, and routing uses your workspace's current settings, so newly added coding agents like Claude Code are picked up.
  • Health tracking now records the specific software environment and version behind a failure, so old failures no longer count against upgraded software. Pinned computers also report their software environment correctly.
  • Long-running Muse Code sessions are no longer ended for inactivity while tools are still running.
  • Fixed an unclassified error that could occur when agent deployments evaluated whether to advance to the next stage.
  • Fixed Google Analytics 4 experiment property IDs, which could cause request failures when they contained special characters.

Roadmap and settings​

  • Roadmap Feedback has a new Requester filter. Include or exclude specific submitters to sort feedback by who asked.
  • Ship maintenance settings are reorganized into collapsible sections, with clearer status indicators and queue tracking.
  • Deployment profiles, maintenance cards, releases, and release watching are now available in every environment.