huggingface_hub v1.13.0 Adds Global CLI Flags for Human and AI Agent Output

huggingface_hub v1.13.0 Adds Global CLI Flags for Human and AI Agent Output

The huggingface_hub library has released v1.13.0, adding three new global flags available on every hf CLI command: --format (output format control), --json (machine-readable output), and -q (quiet mode). The --format auto option is particularly notable: it automatically selects human-friendly output when run in a terminal and agent-friendly structured output when called from AI tools or pipelines.

Why It Matters

Native agent-mode output from the HuggingFace CLI removes friction for AI agents that interact with the Hub programmatically—no more parsing human-formatted text. This positions the HuggingFace CLI as a first-class tool in agentic pipelines alongside programmatic APIs.