hf-spaces

What This Skill Does

The hf-spaces skill empowers OpenClaw to leverage the vast ecosystem of Hugging Face Spaces and Inference Providers for AI-driven creative tasks. It allows users to generate high-quality images, cinematic videos, custom audio, and speech directly from within the agent interface. By utilizing the gradio_client for interacting with hosted Gradio applications and huggingface_hub for direct inference API access, this skill bridges the gap between complex AI models and user-friendly automation. It handles tasks ranging from simple single-model requests to complex chains where output from one Space (like an image generator) serves as input for another (like a video model). Whether you need to batch-generate assets or integrate specialized computer vision, this skill provides the necessary programmatic interface.

Installation

To integrate this skill into your environment, use the OpenClaw command-line interface: clawhub install openclaw/skills/skills/gary149/hf-spaces Ensure you have the required Python dependencies installed in your environment, which include gradio_client and huggingface_hub. Use uv init && uv add gradio_client huggingface_hub to set up your project directory.

Use Cases

• Batch Asset Production: Automate the creation of dozens of variations for marketing campaigns or placeholders using iterative API calls.
• Complex Content Pipelines: Link image generation models with animation models to create dynamic video clips from static inputs.
• Task-Specific AI Selection: Use semantic search capabilities to discover the most appropriate Space for specific requirements like voice cloning or document OCR, rather than hard-coding dependencies.
• Prototyping: Rapidly test different models available on Hugging Face using their Inference Providers before building full-scale applications.

Example Prompts

1. "Generate 10 images of a retro-futuristic city skyline using the Z-Image-Turbo Space and save the metadata."
2. "Search for the best Hugging Face Space for text-to-speech, then convert the following paragraph into an audio file: 'OpenClaw makes agent automation simple.'"
3. "Create a 5-second video from an image of a mountain range; use the ltx-2-TURBO Space to handle the motion generation."

Tips & Limitations

• Rate Limits: When using public Spaces, keep in mind that concurrent requests might trigger rate limits. Implement small delays in your loops if you receive errors.
• Dependency Checks: Always verify if an MCP tool exists for your task before writing custom gradio_client scripts, as MCP tools are optimized for performance.
• Environment Variables: Ensure your HF_TOKEN is set in your environment if you intend to access gated models or private Spaces to ensure uninterrupted service.