Agent-optimized CLI output

Hugging Face described CLI patterns that make tools easier for coding agents to use: compact, complete, parseable, non-interactive output and command references designed for machines. Why it matters: Agent runs often waste tokens and fail on prompts, truncated output, or ambiguous terminal text. Tooling that emits stable structured output can reduce retry loops and improve automation reliability.

Try this: Add a JSON or TSV mode to one script used by agents, document non-interactive flags, and make error output explicit enough for an agent to recover without scraping prose.