Skip to main content
The run_streaming_command() method is designed for executing long-running or interactive commands inside your sandbox while receiving real-time output streams. Unlike run_command(), this method does not block until completion — it allows you to process live logs or incremental output as the command runs.

​
run_streaming_command() -> dict

Description:
Executes a long-running command inside the sandbox and streams output in real time. Callbacks are triggered as new stdout or stderr data chunks arrive.
Parameters:
ParameterTypeRequiredDescription
cmdstrYesCommand to execute (e.g., "npm", "python3")
argsList[str]YesList of command arguments
stdinstrNoOptional standard input to send to the command
on_stdoutCallable[[str], None]NoCallback function invoked when stdout data arrives
on_stderrCallable[[str], None]NoCallback function invoked when stderr data arrives
Returns: 
{
  "stdout": str,   # Complete accumulated output
  "stderr": str,   # Complete accumulated error output
  "exit_code": int # Process exit code (0 = success)
}
Example 

def handle_stdout(data: str):
    print("[stdout]", data.strip())

def handle_stderr(data: str):
    print("[stderr]", data.strip())

result = sandbox.run_streaming_command(
    cmd="npm",
    args=["install", "express"],
    on_stdout=handle_stdout,
    on_stderr=handle_stderr
)

print("Exit code:", result["exit_code"])

## With stdin
result = sandbox.run_streaming_command(
    cmd="python3",
    args=["-c", 'import sys; [print(line.strip()) for line in sys.stdin]'],
    stdin="Hello from stdin\n"
)

print("Output:", result["stdout"])