What problem does it solve? AI coding agents typically run builds, tests, and installs directly on your host machine, contaminating your environment, breaking isolation between projects, and creating security risks. This Skill routes all agent work into isolated dev containers defined by the standard devcontainer spec, keeping your host clean. ## Core Features & Use Cases - Three backends, one interface: Create, manage, and execute commands in dev containers via local Docker (devcontainer CLI), DevPod (multi-cloud: Docker, K8s, AWS, Azure, GCP), or GitHub Codespaces. - Token-safe authentication: An auth broker issues opaque handles for GitHub, AWS, Azure, GCP, and Kubernetes so the agent never sees raw credentials. - Built-in file operations: Read, write, edit, and list files inside any container without constructing shell commands, plus self-healing workflows that let the agent fix a failing Dockerfile or devcontainer.json and retry. - Use Case: Ask your agent to build a Rust project — it spins up a Codespace with your GitHub account, runs cargo build inside it, and stops the environment when done, all without touching your laptop. ## Quick Start Ask your AI agent to start a dev container for the current project and run the test suite inside it using the devcontainer-mcp tools.
