What problem does it solve? Code that passes tests can still be unsafe to deploy: it may lack a revert path, expose unbounded blast radius, ship without metrics, or roll out to 100% of traffic with no kill switch. This Skill reviews PRs through a release-engineering lens so operational risks are caught before merge, not during a 3am page. ## Core Features & Use Cases - Operational readiness checklist: Systematically flags revertability gaps, blast-radius risks, missing observability, unsafe rollout plans, and performance regressions on hot paths. - Build and release path review: Covers Makefiles, Dockerfiles, CI workflows, and release automation that other review lenses skip. - Structured findings: Emits JSONL findings with severity (P0–P3), the check that fired, and—for P0/P1—the mitigating deployment pattern, observability signal, and rollback procedure. - Use Case: A PR adds an endpoint calling a third-party billing API with no timeout, no per-endpoint metrics, and a direct-to-production deploy. The review surfaces three P1 findings (missing timeout, missing metrics, no feature flag) with concrete mitigations before merge. ## Quick Start Ask the AI to review the current branch or PR for operational readiness and deploy safety using the release-engineering checklist.
