Manage Hugging Face models with an AI agent
Use the official Hugging Face CLI to find models, inspect repository information and retrieve a selected file. The same tool also supports authorised uploads and jobs.
To manage Hugging Face models with an AI agent, choose the exact Hub task before handing over a repository name. The hf-cli skill is a reference to the official command-line tool for searching, reading repository details and moving selected files.
For a team comparing two models, downloading a model card first may be enough to settle which checkpoint deserves a trial. Publishing a changed file or starting a cloud job has different consequences. Name the intended account, repository and destination when you move from inspection to an action that writes.
What it does
- Search for candidate model repositories.
- Read model metadata and identify available files.
- Fetch a particular file to a local working folder.
- Upload an approved file to an authorised repository.
- Look up the CLI options for a specific Hub operation.
Install and use
- Install the skill and Hugging Face’s official hf CLI. Authenticate for private repositories or writes, choosing a token with appropriate access.
- Ask “Show the details of these two models and fetch only the model card from the revision I choose.”
- Confirm the local destination. For an upload, deletion or paid job, review the target account and operation before proceeding.
FAQ
Source
Made by Hugging Face. Original skill: hf-cli, published on GitHub under the Apache-2.0 licence.
View on GitHub