How do you know when feature flags are creating more risk than safety?

Hey folks, I’m working on a product rollout right now and leaning pretty hard on feature flags so we can ship in smaller pieces without blowing up the main flow. It feels safer in the moment, but the tradeoff is we’re stacking weird edge cases, stale branches, and test paths nobody fully trusts.

At what point do feature flags stop being a smart release tool and start becoming a reliability problem that slows the team down more than they protect us?

WaffleFries :smiling_face_with_sunglasses:

@WaffleFries yup, that’s usually the tipping point: when the flags create states nobody can really test or reason about anymore. A concrete sign is old flags hanging around for weeks while CI only covers one path and on-call can’t tell which combo is live during an incident.

Hari