Start an experiment. Grow it into a product.
Stario is a small Python framework boiled down to the essentials—explicit routes, handlers, and telemetry that humans and AI agents read the same way. Start an experiment on one stack, stay observable as users try it, and grow without swapping tools or rewriting when it works.
What you get
The core ideas, stated plainly—the same vocabulary for you and your AI agent.
One vocabulary for both
No hidden registration, no magic. Explicit routes, handlers, and responses you and an AI agent can read the same way—so you both focus on what the app actually needs.
One Python codebase
Templates, handlers, server logic, and live UI share one language and one mental model. No second front-end stack.
Live UI when it earns it
Add Datastar and streaming only when the product needs push—the same request model, no new architecture.
Observable by default
Every request, startup, and error is a span that lands in a SQLite file. You don't bolt on observability later.
Read production like code
Query what your live app actually did—slow routes, 5xx, errors—over MCP, so your AI agent can help you fix it.
Grows with your idea
One Stario process behind a reverse proxy comfortably serves thousands of users and hundreds of concurrent connections—far enough to get real customers. Earn your scaling problems instead of pre-paying for them.
Learn by building something real
A counter, a live board, a multi-file app with SQLite—the same patterns that carry an app from first commit to real traffic.
Put your app in orbit. Watch it spin.
Stario Orbit is the operational half: deploy from your repo, observe telemetry while your app runs, and query production through an MCP endpoint your AI agent can reach. Hosting that closes the loop instead of hiding it.