Start from what's already shipped.
Point Statecraft at your code. An agent rebuilds your existing screens and flows on the canvas, with your components and error states.
How it works
From repository to canvas.
Point Statecraft at the code and an agent traces each journey through it. Nobody has to click through the product capturing screens.
- 01
Choose a source.
A repository on GitHub, GitLab, Gitea, Bitbucket or your own server, or a folder on your Mac.
- 02
Survey or target.
Let the agent propose the journeys it finds and tick the ones to build, or describe a single journey.
- 03
Watch it build.
Picked journeys build in parallel. Screens appear on the canvas as the agent adds them.
- 04
Refine.
Give follow-up instructions and the agent edits what it built, without starting again.
What you get
Every state, on one canvas.
The agent follows user actions, API calls and error paths, so the map includes the screens that are hard to reach by clicking.
Error and empty states. Loading, error and empty screens are built alongside the main path.
Your components. Screens are written as React or Vue code with the design system you choose, and every class name is checked against its CSS.
Read-only. The agent reads a shallow clone of your repository and never pushes anything back.
Local folders. Files are served from your Mac as the agent asks for them. Secrets such as .env files are never served.