GeezerVison

_AutoInterrupt now has a timeout attribute and uses it in wait(timeout=...), but no callers pass a value (the only construction is self.AutoInterrupt(proc, command)). As a result, timeout stays None and the new timeout-aware wait()/_terminate() behavior is never applied. If the intent is to enforce kill_after_timeout for as_process=True flows (e.g., via handle_process_output), wire the timeout through when constructing AutoInterrupt or set it before calling _terminate()/wait().

In _terminate(), if proc.wait(timeout=self.timeout) raises TimeoutExpired, the exception is logged and ignored, potentially leaving the child process running and leaked. Consider escalating to proc.kill() (or another stronger termination strategy) after a terminate+wait timeout, and/or ensuring the process is reaped.

execute()'s TimeoutExpired handling calls proc.terminate() and then proc.communicate() without any timeout, which can hang indefinitely if the process ignores termination. Also, status is not updated in this path, so with with_exceptions=False the method may return status=0 even though a timeout occurred. Consider: (1) setting status = proc.returncode after reaping, (2) using a bounded second communicate(timeout=...) followed by proc.kill() as a fallback, and (3) avoiding log text that claims the process was "killed" when only terminate() was attempted.