Hi all,
We discussed a new SIG Apps subproject called agentic-devops in today's call.
As autonomous agents increasingly manage application lifecycles, this subproject aims to develop and maintain tools for agents running DevOps tasks on Kubernetes. Our goal is to provide a framework for verifying and benchmarking how well agents perform operations tasks, helping to establish a "paved path" for production AI operations.
As described in the proposal, our initial artifacts will include two new kubernetes-sigs repos:
devops-bench: A standardized benchmarking suite to evaluate agent performance.kube-agents: A set of reference agents (including relevant MCP server, tools, skills, etc.)Links:
Please take a look at the proposal and let us know if you're interested in collaborating on this effort! I will continue to post updates as we make progress.