A production canary is degraded after a Copilot-generated remediation PR was merged. A release-validation agent detected the regression, and another agent has already opened an incident with the failed workflow run attached. You need to recover service safely before allowing more agent-driven promotion. Which four actions should you perform in order? 1. Ask Copilot to refactor the recovery workflow 2. Cancel the in-progress promotion workflow 3. Rerun all failed agent jobs with debug logging 4. Identify the last successful production deployment SHA 5. Trigger the rollback workflow with that SHA 6. Require human approval before reopening promotion

The correct sequence prioritizes immediate service restoration and then implements safeguards. First, "Cancel the in-progress promotion workflow" (2) stops further deployment of the problematic code, preventing wider impact. Next, "Identify the last successful production deployment SHA" (4) pinpoints a known good state. Then, "Trigger the rollback workflow with that SHA" (5) quickly reverts the system to a stable version, restoring service. Finally, "Require human approval before reopening promotion" (6) adds a critical human gate, preventing automated agents from reintroducing the issue without review and ensuring stability before further automated actions. Option 1 is incorrect because asking Copilot to refactor the recovery workflow (1) is a development task, not an immediate recovery step. Option 3 is incorrect because rerunning failed jobs with debug logging (3) is for diagnosis, not for immediate service restoration. These actions would delay recovery.