Join us as we run a Cursor agent against a real Kubernetes environment and watch it write code, test, and iterate on its own.
Your agent may actually read your codebase or your tests. It does not read the queue payload your producer started emitting last sprint, the database schema that drifted from your migrations, or the downstream service that returns 429 under real load.
Nothing required to attend. These tools will let you adapt the pattern to your stack right after the session.
kind, minikube, or a dev cluster all work).kubectl configured to that cluster.Recently they wrote about how mirrord cut their dev feedback loops from 30 minutes to 30 seconds. Same pattern, different scale.
If you are ready to move past the push-and-pray loop, these 45 minutes are for you.
Save My SpotWho needs AI to finally respect the Kubernetes cluster.
Who needs proof that AI code holds up before it ships.
Who wants agent workflows trusted inside real engineering systems.
Who wants a feedback loop that does not rely on lying mocks.
We are not here to read slides. You are getting engineers with a bias towards action, who build the tools, see diverse use cases in action, and can explain the why.
Senior DevRel Engineer, MetalBear
Arsh has worked in the Cloud Native space for more than 5 years. He will guide the session and connect each step to the developer workflow.
Founder & CEO, MetalBear
Aviram co-created mirrord and leads the MetalBear team. He will explain how AI agents use mirrord and the patterns he has seen work for leading engineering teams.
Register anyway. We will send the recording, slides, and demo repo to your inbox within 24 hours.
The session is 45 minutes, followed by live Q&A. Come for the demo, and stay if you want to dig into your own Kubernetes or agent workflow questions.
No. The session is a live demo, not a hands-on tutorial. Watch what happens, then take the agent instructions, mirrord config, and prompts to try it on your own service after.
mirrord is an open source developer tool from MetalBear. It lets your local process (or AI agent) run in the context of a real Kubernetes cluster: real traffic, real env vars, real file access, no mocks. That is what gives the agent the runtime feedback loop you will see in the session. github.com/metalbear-co/mirrord.
monday.com recently shared how they use a similar workflow across 350+ engineers, cutting dev feedback loops from 30 minutes to 30 seconds. Read their case study.