Kom - Kubernetes Operations Manager

Kom – Kubernetes Operations Manager

Kom is a lightweight, SDK‑style wrapper around the Kubernetes API that exposes a rich set of CRUD operations, advanced pod utilities, and SQL‑style querying for both built‑in resources and Custom Resource Definitions (CRDs). It is designed to be consumed by AI assistants through the Model Context Protocol, enabling developers to orchestrate cluster operations directly from conversational agents without writing code. By registering multiple clusters—whether on‑prem, EKS, or any kubeconfig—the tool provides a single, unified interface for managing resources across environments.

The server’s value lies in its extensive tooling surface. Kom bundles over 58 pre‑built commands, such as scaling or restarting deployments, deleting DaemonSets, and manipulating files inside pods. It also supports high‑frequency operations like restart, scale, enable/disable, and rollback for more than 20 resource types. The ability to query resources with SQL () gives developers a familiar, declarative way to filter and sort cluster data. Additionally, Kom implements caching for list operations, dramatically improving performance in bulk or repetitive queries while still allowing fine‑grained filtering.

Kom’s integration with MCP is straightforward: a single line starts the server, which then exposes two interaction modes—stdio and Server‑Sent Events (SSE). The stdio mode is ideal for command‑line tools such as Claude Desktop, whereas SSE allows UI clients like Cursor or Windsurf to subscribe to real‑time updates. The server’s configuration is declarative, letting users specify whether they want an executable binary or a command‑based integration. This flexibility makes Kom suitable for both scripted automation and interactive workflows.

Real‑world use cases include automated CI/CD pipelines that scale deployments during test runs, on‑demand pod file extraction for debugging, or dynamic cluster provisioning in multi‑tenant environments. Because Kom handles both native Kubernetes objects and CRDs, it is also well suited for operators that manage custom workloads. Developers can embed Kom into their own MCP clients, expose it as a service in a private network, or share it with teammates for collaborative cluster management—all without the overhead of maintaining a full Kubernetes client library.

In summary, Kom provides an opinionated yet extensible MCP server that turns complex cluster operations into simple, repeatable commands. Its SQL querying, caching, and comprehensive toolset make it a powerful ally for AI assistants that need to read, write, and manipulate Kubernetes resources efficiently and securely.