Over 1 million users gain zero-setup access to real-time cluster data for Claude, Copilot, Cursor and ChatGPT.
Platform engineers have spent years building guardrails around Kubernetes while developers chase faster code velocity with AI. Today Mirantis closed one of the last remaining gaps: AI coding assistants can now talk directly to live clusters without exporting kubeconfigs, writing glue scripts, or risking credential leaks.
Lens Desktop, the Kubernetes IDE, is currently being used by more than 1 million developers and operators. It ships a built-in Model Context Protocol server in its 2026.3 release. This is the first major management tool to do so.
The MCP server exposes every cluster already connected in Lens: local kubeconfigs, AWS EKS, Azure AKS, and Lens Teamwork Spaces. Google Cloud support lands soon. AI tools simply discover those clusters through Lens and run read-only operations (get, describe, logs, top, plus PromQL against Prometheus) all while credentials stay on the developer’s desktop.
No custom plugins. No manual context pasting. The protocol, originally open-sourced by Anthropic in late 2024, standardizes secure two-way connections between AI agents and infrastructure data sources. Lens turns that standard into a one-click bridge.
Miska Kaipiainen, head of product for Lens at Mirantis, put it clearly: AI assistants have become daily drivers for developers, yet Kubernetes access stayed awkward. Lens now serves as the secure, native bridge.
Lens Prism moves from diagnosis to action
The same update strengthens Lens Prism, the IDE’s built-in AI assistant. Its new Terminal Skill runs local shell commands in agent mode, letting it generate manifests, apply changes, and push commits directly to version control. And you can do all this while respecting the same security boundaries.
New users get an onboarding wizard that walks through cloud logins and cluster imports, plus a 30-day premium trial. Existing Lens users on Plus, Pro, or Enterprise plans unlock full MCP capabilities immediately.
Mirantis’ long pattern of turning open source into enterprise reality
This release continues a 20-year thread. Mirantis began as a leading OpenStack contributor and operator of some of the largest deployments on the planet. When Kubernetes displaced OpenStack in many enterprises, the company pivoted: it acquired Docker’s enterprise business in 2019, rebranded it as Mirantis Kubernetes Engine, and delivered supported, production-grade distributions.
In 2020 it bought the Lens project from its Finnish creators at Kontena (later Lakend Labs), kept it fully open source under MIT, and grew it from 35,000 users to more than 1 million. Along the way Mirantis released k0s (a lightweight, CNCF-compliant Kubernetes distribution) acquired amazee.io to bring Lagoon, and launched k0rdent in 2025 to simplify multi-cluster AI workloads.
The company’s Open Source Program Office remains active in Kubernetes and CNCF projects while shipping tools that remove operational friction for both developers and platform teams. Embedding MCP follows the same playbook: take an emerging standard, make it native, and keep control and security in the user’s hands.
Real impact on daily Kubernetes workflows
Most teams still lose hours switching between IDEs, terminals, and separate AI chats to answer simple questions like “What pods are using the most CPU across our staging clusters?” or “Show me recent logs for the failing service.” Lens collapses that friction.
Because the MCP server is read-only by design and runs locally, organizations avoid the security risks that come with giving external AI tools broad cluster access. Fixes still route through Lens Prism or GitOps pipelines, which many would propose is exactly where governance belongs.
Platform engineers gain something rarer: AI agents that understand their actual environment, not just yesterday’s YAML. That context gap has widened as generative models push code to production faster than ever.
What this signals for platform engineering
Kubernetes operations have matured, but AI-assisted development is still in its awkward teenage phase. By natively supporting the Model Context Protocol, Lens is positioned as the shared workspace where human developers and AI agents collaborate on real infrastructure, not just code.
The move also highlights a broader industry movement: tools that only manage clusters or only help write code are no longer enough. Practitioners need platforms that connect both worlds without adding complexity or risk.
Lens 2026.3 is available now for download at lenshq.io. The MCP server works with any MCP-compatible AI tool and requires no additional configuration for existing Lens users.
