hfsearch: A fast CLI tool to discover models and datasets on HuggingFace (github.com)

hfsearch is a new lightweight Python library and command‑line tool that makes discovering models and datasets on the Hugging Face Hub fast and scriptable. Installable via pip (pip install hfsearch) or editable from source, it exposes a simple Python API (search_models, search_datasets, export_to_csv) and a CLI (hfsearch models|datasets) to query the Hub with filters for query, limit, author, tags and task. Results are returned as lists of dictionaries (id, author, downloads, likes, tags) and can be exported to CSV or TXT; terminal output is prettified with Rich. For practitioners, hfsearch speeds model/dataset discovery and automation in preprocessing, benchmarking or CI workflows by enabling targeted searches and bulk exports without manual browsing. It relies on the Hugging Face Hub API (huggingface_hub>=0.20.0) and requires Python 3.7+; Rich (>=13.0.0) is used for formatted output. Authentication is optional but recommended for private assets, and hfsearch intentionally does not download model files (use the official huggingface-cli for downloads). The project is MIT‑licensed and open to contributions, making it a handy, extensible helper for ML engineers, researchers, and devs who need quick, filterable access to Hub metadata.