Docker launches hosted sandboxes for AI coding agents
Developers can shift agent work off a laptop without losing its filesystem state, though sandboxing does not limit what agents can do through services they can access.
Docker has launched Cloud Sandboxes, hosted environments for AI coding agents that bring its local sandbox approach to Docker-managed infrastructure, with hardware-enforced microVM isolation. Developers can shift a sandbox between a laptop and the cloud using `sbx move my-project --to cloud`; Docker says the transfer preserves the filesystem and supports multiple agents running for hours or working in parallel. The company says agents can run for five, ten or 21 hours each, without developers monitoring them. Docker also released Kits under its v3 specification as standard OCI images, usable with image operations such as build and pull and as starting points for more complex Kits. Commenters warned that isolation does not govern an agent’s interactions with services it is allowed to access, and one argued that capability-based controls could offer a better approach.
Why it matters
For developers, cloud hosting makes it possible to continue agent work beyond the local environment without having to carry over the project state separately. But containing an agent does not resolve the risk of its interactions with services it can reach.
Matters or hype?
Does this story matter, or is it hype? Vote before you see what everyone else said.