Skip to main content

Introduction

k8shell is a platform built around a simple idea: developer workspaces run in Kubernetes. Developers can have direct access to backends, databases, and internal services - the same resources available to apps in test or production.

For AI agents, k8shell workspaces provide a natural isolation boundary. Agents run with no credentials on disk, no access to the host filesystem, and workspace configuration restricted by OPA policy rules. Using the k8shell MCP server, AI agents can spawn workspaces on demand and perform their tasks in a fully isolated, auditable environment — with access scoped to exactly what the workspace allows.

A workspace is provisioned on demand and accessible over SSH, a browser-based Console, or API.

Prefer learning by doing?

Follow the Quick Start to get a working installation up in minutes and explore k8shell hands-on.

How the docs are organized

Introduction — covers the open-source license, Early Access program, how to get help, and roadmap.

Overview — explains how k8shell works: the architecture, core services, security model, and workspace lifecycle. Start here to understand the platform before going deeper.

Services — documentation for each k8shell service and how they interact:

Installation — step-by-step guides for deploying k8shell.

Configuration — Helm Charts reference and guides for configuring k8shell.

User Guide — How to connect to and work within k8shell workspaces via Console, CLI, and SSH.