Piping CLI output through tail or head can hang forever if another process holds the write end

A developer spent weeks misdiagnosing a vendor CLI as unreliable, only to discover the real culprit was piping output through tail or head commands. Both utilities wait for EOF before producing output, but EOF only arrives when all processes holding the pipe's write end have closed it. In multiple incidents, unrelated background processes — such as npm exec xcodemcp@latest and npm exec @modelcontextprotocol/server-redis — had inherited the pipe's write end and kept it open indefinitely. A shell-level alarm timeout also failed to help because the exec call replaced the original process, leaving the signal with no valid target. The fix is to redirect output fully to a file first, then apply tail or head to that file, or use a runtime like Python's subprocess.run with capture_output=True to avoid shared pipe descriptors entirely.