Quick Start

1. Install & sign in

curl -fsSL https://github.com/Podstack-ai/podstack-cli-releases/releases/latest/download/install.sh | sh
podstack auth login          # opens your browser (Google / GitHub / SSO)

login also selects your default project. If you have more than one, pick it later with podstack projects use <name>.

2. Build and preview an app

mkdir my-app && cd my-app
podstack code

At the prompt, describe what you want — for example:

“Build a FastAPI todo API with a small React front end, then show me a live preview.”

The agent recommends the app shape, plans it, installs everything in a cloud sandbox, builds it, and hands you a public preview URL. Backends include interactive Swagger docs, and every project gets a README.md.

Track spend with /cost, and stop the preview (and its billing) with /preview off. See Coding agent.

3. (Optional) Rent a GPU

podstack gpu launch                 # pick a GPU and launch it
podstack gpu instances ssh <id>     # connect over SSH

See GPUs and the TrainPod guides.

4. (Optional) Call a model

podstack models list                # models on Inference Cloud

Use them over an OpenAI-compatible API — see Inference.

Where to next

• Coding agent — the full podstack code reference.
• Sandboxes — drive previews directly.
• FAQs & scenarios — common questions and end-to-end walkthroughs.