Runners Controller 0.5.0 introduces account-scoped Runner Profiles for Linux. You can now define named profiles that bundle vCPU and RAM, then reference them in GitHub Actions with runs-on: tuist-<name>. Updating a profile’s resources applies to every workflow that uses it, so you can scale runners without editing workflow files.
- Manage profiles from the new Runners / Profiles dashboard. The label shown in the UI is the same name you pass to
runs-on. - Linux-only in this version; macOS profiles will follow. Each account can create up to 10 profiles.
- Existing workflows using the legacy
tuist-{env}-linuxlabel continue to run without changes and are mapped to the default 4 vCPU / 16 GB shape, matching GitHub’s standard Ubuntu runner. - The fleet scaler is now capacity-aware across heterogeneous shapes: shared bare-metal capacity is rebalanced between warm buffers and real work, so one idle runner type does not starve another’s jobs.
- The runners Grafana dashboard adds a capacity-fit panel and an unscheduled-replicas stat to help track fragmentation and provisioning needs.