I would use this as a human. That npm output is crazy. Maybe a better variable would be "CONCISE=1". For LLMs, there are a few easier solutions, like outputing in a file (and then tail)., or running a subagent