Creates, deploys, debugs, and maintains apps on Hugging Face Spaces across Gradio, Docker, and Static SDKs - choosing hardware tiers (cpu-basic, ZeroGPU, dedicated GPU), model sourcing, and grants.
---
name: huggingface-spaces
description: Build, deploy, and maintain applications on Hugging Face Spaces - Gradio / Docker / Static SDKs, ZeroGPU and dedicated hardware, model loading, debugging, and community grants. Use to create or host an app on HF, port code onto ZeroGPU, or fix a Space that won't build or run.
---
# Hugging Face Spaces
Each Space is a git repo hosting an ML app. This skill covers creating, building, debugging, and maintaining them.
## Workflow
1. Get ready: confirm `hf` is installed and the user is logged in (`hf auth whoami`); note `canPay`/`isPro` flags, which gate hardware.
2. Search for prior art first: `hf spaces search "<model or task>" --sdk gradio` and read a working `app.py`/`requirements.txt`.
3. Pick the SDK - Gradio (default, supports ZeroGPU), Docker (non-Python stacks), or Static (in-browser ML, project pages).
4. Pick hardware - `cpu-basic` (free) or ZeroGPU `zero-a10g` (free for creator, PRO-gated, Gradio-only) by default; dedicated GPU (T4/L4/A10G/A100/H200) only when ZeroGPU doesn't fit.
5. Source the model (GitHub repo, HF model repo, or Inference Providers), estimate VRAM, write the app, and push.
6. For ZeroGPU on a non-PRO account, build on cpu-basic and request a community grant.
Full skill & source: https://github.com/huggingface/skills/tree/ea9a24f38a98baf2430dab3bcb31a89753ae9e6d/skills/huggingface-spaces