Trust & data handling
Your files. Your machine. No cloud.
Scamp is a local-first desktop application. It reads and writes plain files inside a project folder you control. It does not upload your designs, does not send telemetry about your layouts, and has no backend for your work to live in.
Quick summary
| Concern | How Scamp handles it |
|---|---|
| Where designs are stored | On your local disk, in the project folder you choose. |
| Cloud sync | None. Scamp has no servers for design data. |
| Account | Not required to use the app. |
| Telemetry | No design data is collected. |
| File access scope | The project folder you open. Nothing outside it. |
| Network calls | Only when checking for app updates or when you explicitly open a link. |
| Lock-in | None. Projects are plain TSX and CSS Module files you can keep using without Scamp. |
What Scamp reads
When you open a project, Scamp reads the files inside that project folder so it can render them on the canvas:
- Page files — the
.tsxand matching.module.cssfiles that describe each page layout. - theme.css — your design tokens (colors, spacing, radii, fonts) defined as CSS custom properties.
- agent.md — an instructions file that AI coding agents read so they can edit Scamp projects correctly.
- Image assets — files you reference in your pages (e.g. PNGs, SVGs, user uploads) so the canvas can render them.
Scamp does not read files outside the project folder. Opening a different project means opening a different folder — nothing crosses over.
What Scamp writes
Scamp writes changes back to the same files you opened. Every edit on the canvas is a change to real code on disk — not a database entry somewhere.
- Editing a layout writes the updated TSX and CSS Module files for that page.
- Editing a theme token writes
theme.css. - Renaming a page renames the corresponding files.
- App-level preferences (default project folder, theme) live in your OS user config directory, not in your project folder.
What Scamp does not do
- No cloud sync. Your designs never leave your machine unless you explicitly push them to a git remote or send the files yourself.
- No analytics on your designs. The shapes, colors, and layouts you draw are not reported anywhere.
- No account, no login. Scamp does not have a user directory for you to exist in.
- No plugin marketplace execution. Scamp does not auto-run third-party code from an online marketplace.
AI coding agents & permissions
Scamp has a built-in terminal that opens in your project folder. You can run any CLI coding agent — Claude Code, Codex, Gemini CLI, OpenCode, or whatever else you use — the same way you would run it in a normal terminal.
- The agent runs under your user account. Its file access is whatever your shell already allows — Scamp does not grant extra permissions.
- Approvals are controlled by the agent itself. If the agent requires human approval for destructive actions, Scamp inherits that behavior.
- Scamp reads
agent.mdfor instructions. This file tells the agent how Scamp expects its project files to look. You can read, edit, or delete it. - You can turn it off. The terminal panel is optional. If you never open it, no agent ever runs.
Reporting a concern
Think you found a security issue or a data-handling bug? Get in touch and we will triage it as soon as it comes in.
Want to understand how Scamp works end-to-end? Start with Bidirectional Sync in the docs.