How it works

1. 01
   Your agent reads the folderIt sends the source files — not node_modules, not .git, not your .env.
2. 02
   We bundle and store themThe files become a versioned archive. Every version is kept, so any of them can be restored later.
3. 03
   A container startsIsolated, with a fixed memory ceiling and its own filesystem. Nothing it does can affect another customer.
4. 04
   Dependencies installFrom your package.json or requirements.txt. Install scripts are disabled — a package cannot run arbitrary code at install time.
5. 05
   Your entry point runsOutput is captured from the first line, so an app that fails immediately still tells you why.

A crashed process is restarted automatically. If it crashes repeatedly — five times in half an hour — we stop retrying and tell you, because a restart loop hides the actual problem instead of solving it.

If the machine your app is on fails, it is rebuilt somewhere healthy from its last deploy. You do not need to do anything.

• We store your code so it can run, and keep earlier versions so you can roll back.
• We read your logs when you ask us to help with something, or when abuse is detected. Not otherwise.
• Environment values are encrypted and write-only. Nobody reads them back — not you, not us, not a connected agent.
• We do not train anything on your code.