Keptn is a Kubernetes-native tool for evaluating deployments against Service Level Objectives and running pre- and post-deployment checks. In its current incarnation (Keptn Lifecycle Toolkit) it runs as an operator that watches Deployments, StatefulSets, and DaemonSets and attaches KeptnTask and KeptnEvaluation resources to their rollout lifecycle.
The core idea is the KeptnMetric and KeptnEvaluationDefinition CRDs: you declare SLOs as queries against Prometheus, Datadog, Dynatrace, or any provider implementing the metrics provider interface, and Keptn blocks or fails a rollout if the query result is outside the expected range. It emits OpenTelemetry traces spanning from “application version deployed” through every pre/post task so you can see deployment lifecycle data in the same backend as your runtime traces.
The original v1 Keptn was a heavier event-driven delivery platform built around its own Shipyard/Cloud Events control plane; that version was superseded by the lightweight lifecycle toolkit. Keptn was archived by the CNCF in September 2025, so new adopters should expect to be on their own for maintenance — the functional overlap is with Argo Rollouts analysis templates and Flagger for the same SLO-gated rollout use case.