Find the user's Box environments and inspect their current lifecycle states.

Inspect one Box environment after create, stop, resume, or fork operations.

Provision a paid Box Linux environment owned by the configured Box account. This changes external state and starts billable runtime.

Stop a Box and snapshot its disk. force=true may irreversibly lose changes since the last successful snapshot and should only follow a refused normal stop.

Resume an archived Box, restarting billable runtime on a fresh machine.

Create a new paid Box from an existing snapshot. The source Box is not modified.

Run a shell command inside a Box and wait up to 60 seconds. The command may modify files, install software, access configured credentials, or affect external systems. Use box_job_start for longer work.

Read a file relative to the Box work directory. Results are capped for model safety.

Create or replace a file relative to the Box work directory. Existing content at the path is overwritten.

Inspect recent lifecycle and work events emitted by a Box.

Create a secret-bearing desktop streaming URL for the user to inspect the Box. Do not log, persist, or share the returned URL.

Start a long-running shell job in a Box. The job runs independently after this tool returns and may modify the machine or external systems.

Check whether a background job is running, succeeded, failed, or was cancelled.

Read the tails of stdout and stderr for a background job.

Send SIGTERM to a running background job and its process group. Partial file or external changes are not rolled back.