kind, metadata,
spec, and read-only status fields, and it uses a pod template to describe
the Linux environment to run.
When to use a Sandbox
Use a Sandbox when you need a temporary environment for:- Agent workspaces.
- Experiments.
- Tests.
- Builds.
- Debugging.
- Preview environments.
Lifecycle
Every Sandbox has a requiredspec.ttl. The TTL uses duration strings such as
15m, 2h, or 24h. IDYL removes the Sandbox after the duration expires
unless it is deleted earlier.
Use idyl delete sandbox <name> to remove a Sandbox before expiry.
Names and placement
The imperative CLI workflow can generate a Sandbox name:--name when you want to choose the name:
metadata.name:
metadata.subnet in manifests or --subnet in CLI commands to choose the
subnet where the Sandbox should run.
Isolation
Sandboxes can request a workload isolation class when allowed by the target subnet:
The isolation class is part of the Sandbox pod template. The target subnet and
compatible provider capacity must admit the requested value.
For the broader isolation model, see
Workload isolation.
Runtime status
When available, Sandbox status reports the associated pod for the environment. Inspect the Sandbox to see TTL, expiry, image, and associated pod information:idyl get sandboxes to list Sandboxes in the current namespace.
Command scope
The documented Sandbox CLI workflow is create, inspect, apply, and delete:shell, exec, cp, logs, wait, and scale commands
are not part of the documented Sandbox CLI workflow. Use the published workload
commands for behavior documented outside the Sandbox workflow.

