Q: Your team frequently accidentally breaks the `staging` environment because developers merge half-finished features under feature flags, but the flags are never cleaned up. Staging accumulates stale flag state and begins behaving differently from production. How do you bring hygiene to feature flag management in your CI/CD process?

Feature flag rot is a process problem, not just a tooling one:

• Expiry dates on flags: Require every flag created via your flag management tool (LaunchDarkly, Unleash) to have an explicit expiry date. After that date, the platform automatically disables the flag and creates a Jira ticket to remove the code.
• Flag-removal PR as part of the release process: Add a pipeline step that checks for flags that have been at 100% rollout for >7 days and opens an automated PR removing the flag and its conditional branch.
• Staging flag parity check: Add a CI job that compares the flags enabled in staging vs production. Any flag enabled in staging but not production that is older than 2 weeks triggers a Slack alert to the owning team.

• Clean staging on demand: Script a monthly staging environment reset — tear down and recreate from the infrastructure-as-code template, getting a clean flag state matching the IaC defaults.