Release Workflow Kill Switches

Purpose

Kill switches give maintainers an emergency brake for every automated release channel without having to edit .github/workflows/release.yml. Setting the appropriate repository variable forces the corresponding job to short-circuit before any tooling runs, which is safer than letting a known-bad build progress and fail later in the pipeline.

How the guard works

1. Each job starts with a PowerShell probe

Every release job begins with a Kill Switch Check step that reads a repository variable, writes a warning when the switch is active, and emits an enabled output. When the switch is off (false or undefined), the step prints a pass message and sets enabled=true so the remainder of the job can execute.

- name: Release Kill Switch Check
  id: release_kill_switch
  shell: pwsh
  run: |
    $disabled = '${{ vars.RELEASE_DISABLED }}'
    if ($disabled -eq 'true') {
      Write-Host "::warning:: Release workflow is currently DISABLED via kill switch (RELEASE_DISABLED=true)"
      echo "enabled=false" >> $env:GITHUB_OUTPUT
    } else {
      echo "enabled=true" >> $env:GITHUB_OUTPUT
    }

2. All downstream steps are gated

Every subsequent action in the job checks the enabled output. If the switch is active the expression evaluates to false, GitHub skips the step, and the build/page/publish stages never start.

- name: Build Release
  if: steps.release_kill_switch.outputs.enabled == 'true'
  run: msbuild "Scripts/Build/build.proj" /t:Build /p:Configuration=Release /p:Platform="Any CPU"
- name: Push NuGet Packages to nuget.org
  if: steps.release_kill_switch.outputs.enabled == 'true'
  id: push_nuget
  shell: pwsh

Configuration

Branch / Job Id Variable name Protected artifacts
master (release-master) RELEASE_DISABLED Stable builds, packages, Discord master announcement
V105-LTS (release-v105-lts) RELEASE_DISABLED Latest LTS stream (shares the same switch to avoid drift)
V85-LTS (release-v85-lts) LTS_DISABLED Legacy LTS channel
canary (release-canary) CANARY_DISABLED Canary builds and related Discord/webhook posts
alpha (release-alpha) NIGHTLY_DISABLED Nightly/alpha drops and notifications

Why two LTS jobs share RELEASE_DISABLED: the V105 branch follows the same release cadence as master, so maintainers typically want to freeze both at once. If you need V105 to keep running while master is frozen, temporarily duplicate the kill-switch step under a new variable before toggling.

Setting or clearing a switch

  1. Open the repository in GitHub and navigate to Settings → Secrets and variables → Actions → Variables.
  2. Select the variable listed above or create it if it does not exist.
  3. Set the value to the string true to disable that job. Leave it empty or set to false to allow runs. Please note that these values are case sensitive and must always be in lowercase.
  4. Save. The next workflow run on the matching branch will respect the new state without requiring a commit.

Verifying the state during a run

  • When disabled, the job log shows the warning emitted by the PowerShell step and the job exits with a green “Skipped” badge because all guarded steps evaluate to false.
  • When enabled, the log includes “kill switch check has passed” and normal build/pack/publish steps begin immediately afterward.

Operational playbook

Freezing a channel before a risky commit

  1. Flip the relevant variable to true.
  2. Push your work—CI will run but the guarded job will skip, preventing packages and notifications.
  3. Once the branch is healthy again, set the variable back to false and rerun the latest workflow from the Actions tab if you need artifacts.

Recovering after an accidental freeze

  1. Confirm the variable value in Settings.
  2. Re-enable (false), then re-run the failed workflow from Actions → select the skipped run → “Re-run all jobs”. Because kill switches act at runtime, no workflow edits are required.

Auditing who changed a switch

Repository variables track the last editor in the activity log (Settings → Audit log). Filter for repo.variable.update events to see when and by whom a kill switch was toggled.

Extending the pattern

To add a new kill switch (e.g., for a future beta branch), copy the existing pattern:

  1. Add a Kill Switch Check step with a unique id and variable name.
  2. Gate every subsequent step with if: steps.<id>.outputs.enabled == 'true'.
  3. Document the variable under Configuration so on-call engineers know which switch to flip.

Keeping the guard self-contained inside the workflow ensures you can pause or resume publishing safely even while other infrastructure (NuGet, Discord, etc.) stays configured.