Building KubeAgent: an AI on-call assistant for Kubernetes
Hey Product Hunt folks,
I'm building KubeAgent, a CLI for Kubernetes teams who are tired of finding out about cluster issues only after something is already on fire.
The idea is simple: run `kubeagent onboard` once to scan your cluster and build a local knowledge base, then run `kubeagent watch`. When something looks wrong, KubeAgent checks logs/events/describes, tries to explain the likely root cause, and suggests a fix. Safe actions like pod restarts can be applied automatically; risky actions wait for human approval through Slack, PagerDuty, Discord, Teams, Telegram, or a webhook.
I'm building it because I wanted something between "another alert in a dashboard" and "give an AI agent full access to prod." The hard part has been drawing the safety line clearly enough that it can be useful without being reckless.
Current focus:
- better Kubernetes issue detection
- clearer incident writeups
- safer approval flows
- making onboarding feel boring in a good way
Would genuinely appreciate feedback from anyone running Kubernetes in a small team or solo setup. What would make you trust, or not trust, an AI-assisted on-call tool?
Site: https://kubeagent.net
Replies
This sounds like a thoughtful approach to AI-assisted Kubernetes monitoring without going overboard on autonomy. Building in those safety checks and approval flows seems key for real-world trust. If you're up for it, I'm launching on PH soon... would appreciate a follow (See "PRODUCT HUNT LAUNCH" Link in my profile). The Sponge is an AI flashcard tool that turns any content into spaced repetition study material.
Retime
@rianbrob cool. would you use it though Rian? if not, why?