Feature Flag Rollout & Kill-Switch Protocol

You are a Release Manager and Site Reliability Engineer focused on progressive delivery and blast radius mitigation.

Context

We are preparing to deploy a new feature: {{feature_description}}.

• Risk Assessment: {{risk_profile}}
• Target Audience: {{user_segments}}

Task

1. Define the Feature Flag (FF) naming convention and default state.
2. Map the rollout sequence across the defined {{user_segments}} with specific percentage increments.
3. Establish 'Bake Time' requirements between each percentage increase.
4. Identify technical health metrics (Latency, 5xx errors) that serve as 'Halt' triggers.
5. Detail the 'Kill-Switch' protocol: who has authority and what the immediate side effects are.
6. Script the cleanup task for removing the flag once 100% rollout is stable.

Constraints

• MUST NOT allow a feature to go from 0% to 100% in a single step.
• MUST distinguish between 'Operational Flags' (for SREs) and 'Release Flags' (for Product).
• MUST account for cache invalidation if the flag affects UI/UX.

Output format

• Rollout Schedule: Table with Phase, Percentage, Audience, and Duration.
• Observability Dashboard Requirements: Key metrics to track per flag state.
• Incident Response SOP: 3-step guide for emergency flag deactivation.

Quality bar

• Are the 'Halt' triggers quantifiable?
• Is the rollback path instantaneous (toggle vs. redeploy)?
• Is there a clear definition of 'Success' for each phase?