Skip to main content
Apps that handle files — avatars, receipts, CSV exports, media — get real object storage (Amazon S3 underneath). Uploads and downloads work the way production apps do it: your app mints a secure, expiring link, and the user’s browser transfers the file directly to storage. Large files never squeeze through your app’s own server.

​
What this means for your app

  • Users can upload files of any practical size; downloads are served through expiring signed links, so files aren’t publicly listable.
  • Files persist independently of the app — they survive restarts, sleeps, and new deploys.
  • The agent can list, inspect, and clean up stored files while building (“delete the test uploads”).

​
Isolation

Every file your project stores lives under your project’s own storage scope. The scoping is applied at the platform boundary — your app physically cannot read or write another project’s files, and deleting a project’s files affects that project alone.

​
Good to know

  • Signed links expire (15 minutes by default, configurable up to 7 days) — this is a feature, not a bug: a leaked link goes stale.
  • Storage is for files. API keys and credentials belong in Secrets, which is encrypted and rotatable — the agent follows this rule and so should any code you add after exporting.
  • Your app never holds cloud credentials; it requests links through the platform’s runtime broker. That’s also why the same code works identically in preview and production.