hf-mcp

What it does

This skill allows an agent to interact with the Hugging Face Hub using its Model Compute Pipeline (MCP) server tools. It enables searching for models, datasets, Spaces, and papers hosted on the Hub. The agent can also retrieve repository details, access documentation, initiate compute jobs, and utilize Gradio Spaces as integrated AI tools.

When to use it

• Researching new NLP models: An agent could be tasked with finding state-of-the-art sentiment analysis models available on the Hugging Face Hub.
• Finding datasets for a specific task: The agent can locate suitable datasets, such as image classification or text summarization, hosted on the Hub.
• Integrating interactive demos: An agent could use Gradio Spaces to provide users with an interactive interface for testing different models.
• Automating compute tasks: The agent could be used to initiate and monitor training jobs or inference pipelines using resources available through Hugging Face's infrastructure.

Key capabilities

• Search for models, datasets, Spaces, and papers on the Hugging Face Hub.
• Retrieve repository details (e.g., commit history, file structure).
• Access documentation associated with Hugging Face resources.
• Run compute jobs on the Hugging Face Hub.
• Utilize Gradio Spaces as AI tools.

Example prompts

• "Find me the top 5 most downloaded text classification models on the Hugging Face Hub."
• "What is the documentation for the 'bert-base-uncased' model?"
• "Launch a compute job to fine-tune this model using this dataset."

Tips & gotchas

• This skill requires an active connection to the Hugging Face Hub via MCP server tools. Without that connection, it will not function.