Hugging Face
Inspect models, datasets, and Spaces on Hugging Face. Search for pre-trained models, read model cards, compare metrics, and explore datasets — all from within Codex.
Install
npx codex-marketplace add openai/plugins/plugins/hugging-face --plugin Hugging Face is an official OpenAI Codex plugin available through the Codex Marketplace. Install it with one command and Codex gains the ability to work with Hugging Face directly from your coding session.
Install:
npx codex-marketplace add openai/plugins/plugins/hugging-face --pluginThen use it naturally: codex "Hugging Face — describe what you want to do".
The plugin follows the official OpenAI plugin spec with a plugin.json manifest, SKILL.md files, and marketplace.json integration. The source lives in the openai/plugins GitHub repository (893 stars).
Features
Model search
Find pre-trained models by task, architecture, and performance metrics.
Model card reading
Read full model cards including intended use, limitations, and benchmark results.
Dataset exploration
Browse dataset cards, view sample records, and check dataset sizes.
Common Use Cases
- Finding the right pre-trained model for a machine learning task
- Comparing model benchmark scores before choosing one
- Researching training datasets for fine-tuning projects