# Kubernetes MCP servers

MCP servers that let an agent inspect and manage Kubernetes clusters: pods, deployments, logs and events.

## Before you choose

Most Kubernetes servers run locally with your kubeconfig, so the agent gets the permissions of that context; managed cloud servers sign in with the provider.

Point the agent at a context bound to a read-only role first, and give write access only for a specific namespace.

## Servers (20)

What search for "kubernetes" returns, without listings scored poor, published under many names from one repository, or with nothing to connect to.

- [com.googleapis.container/mcp](https://protogrid.dev/servers/com.googleapis.container/mcp.md) `com.googleapis.container/mcp`: class R0, quality 91 (good), trust 69. Provides read access to your GKE and Kubernetes resources.
- [io.github.dockndevai/mcp-kubernetes](https://protogrid.dev/servers/io.github.dockndevai/mcp-kubernetes.md) `io.github.dockndevai/mcp-kubernetes`: class L0, quality – (not scored), trust 91. Kubernetes monitoring & ops for AI agents — safe-by-default access modes and guards.
- [io.github.containers/kubernetes-mcp-server](https://protogrid.dev/servers/io.github.containers/kubernetes-mcp-server.md) `io.github.containers/kubernetes-mcp-server`: class L0, quality – (not scored), trust 86. A Model Context Protocol (MCP) server for Kubernetes and OpenShift
- [Terradev](https://protogrid.dev/servers/io.github.theoddden/terradev.md) `io.github.theoddden/terradev`: class R0, quality 57 (needs work), trust 90. Cross-cloud GPU orchestration CLI.
- [OneUptime](https://protogrid.dev/servers/com.oneuptime/mcp.md) `com.oneuptime/mcp`: class R0, quality 69 (new), trust 100. Manage incidents, alerts, monitors, status pages, on-call and telemetry in OneUptime.
- [Utility Matrix](https://protogrid.dev/servers/io.github.helphub369/utility-matrix.md) `io.github.helphub369/utility-matrix`: class R0, quality 75 (good), trust 69. Free browser-based calculators and analyzers for cloud cost, DevOps, security, and data governance.
- [Kubernetes Troubleshoot](https://protogrid.dev/servers/io.github.NanaGyamfiPrempeh30/k8s-troubleshoot-mcp.md) `io.github.NanaGyamfiPrempeh30/k8s-troubleshoot-mcp`: class L0, quality – (not scored), trust 90. Read-only Kubernetes diagnostics: pods, logs, events, workloads, services, PVCs and nodes.
- [Kubernetes Manifest Audit](https://protogrid.dev/servers/io.github.UnbearableDev/k8s-manifest-audit.md) `io.github.UnbearableDev/k8s-manifest-audit`: class R2, quality 67 (needs work), trust 87. kube-linter audit for Kubernetes manifests — 63 checks: security, availability, RBAC, network.
- [io.github.apatilgtn/k8scortex-mcp](https://protogrid.dev/servers/io.github.apatilgtn/k8scortex-mcp.md) `io.github.apatilgtn/k8scortex-mcp`: class L0, quality – (not scored), trust 81. Kubernetes MCP server: 75 tools for GKE, AKS, EKS, RBAC, audit logging, GitOps, multi-cluster.
- [MimiOps mcp server for Kubernetes](https://protogrid.dev/servers/io.github.sergelogvinov/mimiops-mcp.md) `io.github.sergelogvinov/mimiops-mcp`: class L0, quality – (not scored), trust 94. MimiOps is an opinionated MCP server for Kubernetes
- [RationalBloks](https://protogrid.dev/servers/com.rationalbloks/mcp.md) `com.rationalbloks/mcp`: class R0, quality 70 (needs work), trust 97. Deploy production REST and Neo4j Graph APIs from JSON schemas in seconds.
- [io.github.oijusti/k8s-port-forward-mcp](https://protogrid.dev/servers/io.github.oijusti/k8s-port-forward-mcp.md) `io.github.oijusti/k8s-port-forward-mcp`: class L0, quality – (not scored), trust 76. Kubernetes Port Forward — MCP Server: port-forward Kubernetes services with terminal integration
- [io.github.kubeleash/kubeleash](https://protogrid.dev/servers/io.github.kubeleash/kubeleash.md) `io.github.kubeleash/kubeleash`: class L0, quality – (not scored), trust 80. Kubernetes MCP server with RBAC-style, context-scoped guardrails for AI agents.
- [io.github.mikhae1/kubeview](https://protogrid.dev/servers/io.github.mikhae1/kubeview.md) `io.github.mikhae1/kubeview`: class L0, quality – (not scored), trust 96. Read-only Model Context Protocol MCP server enabling code-driven AI analysis of Kubernetes clusters.
- [Kubernetes Ephemeral Job](https://protogrid.dev/servers/io.github.inhuman/mcp-k8s-ephemeral-job.md) `io.github.inhuman/mcp-k8s-ephemeral-job`: class L0, quality – (not scored), trust 80. Runs a command in a throwaway Kubernetes pod and returns exit code, output and artifacts.
- [io.github.dockndevai/mcp-openshift](https://protogrid.dev/servers/io.github.dockndevai/mcp-openshift.md) `io.github.dockndevai/mcp-openshift`: class L0, quality – (not scored), trust 91. Safe-by-default MCP for OpenShift / Kubernetes: projects, pods, logs, deployments, routes.
- [io.github.rex/rancher-mcp](https://protogrid.dev/servers/io.github.rex/rancher-mcp.md) `io.github.rex/rancher-mcp`: class L0, quality – (not scored), trust 87. Capability-aware MCP server for Rancher-managed Kubernetes: discovery, resources, workflows.
- [io.github.kubently/kubently](https://protogrid.dev/servers/io.github.kubently/kubently.md) `io.github.kubently/kubently`: class L0, quality – (not scored), trust 81. Troubleshoot Kubernetes agentically: natural-language cluster diagnosis via ask_kubently
- [K8gentS — Kubernetes RCA MCP Server](https://protogrid.dev/servers/io.github.JDoornink/k8gents.md) `io.github.JDoornink/k8gents`: class L0, quality – (not scored), trust 80. Kubernetes RCA agent that sandboxes its own AI with OPA Gatekeeper. MCP-ready for any client.
- [io.github.nogoo9/no-crd](https://protogrid.dev/servers/io.github.nogoo9/no-crd.md) `io.github.nogoo9/no-crd`: class L0, quality – (not scored), trust 87. Dynamic pod spawner & proxy for ephemeral AI agent workspaces on Kubernetes without CRDs

All use cases: https://protogrid.dev/mcp.md