Watch the command as it runs.

Run a command in a sandbox and read stdout and stderr as they happen, then the exit code, from the CLI, the API or your agent.

Sandboxes › Live output

Live output

Run a command in a sandbox and read stdout and stderr as they happen, then the exit code, from the CLI, the API or your agent.

sandscale exec sbx_9a14 -- pytest -qstreaming
01........................................ [ 96%]
02FAILED tests/test_charge.py::test_refund - AssertionError
03.. [100%]
041 failed, 41 passed in 3.4s
05exit 1
Streams
stdout and stderr, separately
API
Streamed as it is produced
History
Each run is kept with its command and exit code
Logs
Run output is searchable afterwards

How it works

Three steps, the same from the API, the CLI or your agent over MCP.

01Run

Send the command; it starts in the sandbox straight away.

02Stream

Output arrives line by line, stdout and stderr kept apart.

03Finish

You get the exit code, and the run is in the sandbox’s history.

Which plans

Live output works on every plan.

See pricing →

Give every agent
a box of its own

Start free with 100 vCPU-hours a month. Upgrade when your agents need more.

A Firecracker microVM for every runNo network by defaultPause and resume with state keptVault keeps secrets out of promptsSSO, roles and audit logBring your own cloudWorks over MCP and the CLI