Hi, thanks for asking. Here is exactly how it works.
What is collected (on by default, you can turn it off):
- Anonymous usage events (screens opened, features used), with project identifiers hashed locally before sending.
- A downsized screenshot of the app window when you close it.
- The questions you ask AgentsRoom Help and its answers.
- Hostname and project names are never sent in clear.
What never leaves your machine, whatever the settings: your code, file contents, prompts, terminal output and project names in clear.
How to turn it off: Settings > Privacy & data > switch off "Share usage data". Once off, nothing from the list above is sent.
Two other things to know in that same screen:
- "Automatic AI error diagnosis": when an agent CLI fails to start, the end of the failing terminal output and the launch command (secrets redacted locally) can be sent to an AI service to suggest a fix. Switch it off if you want this to happen only when you click.
- Voice dictation and smart model selection send only the minimum needed to run that feature when you use it, even with sharing off. Don't use them if you want zero exchange.
The community leaderboards and member contact options are also in that screen, and you can leave them entirely.
Full details are in the privacy policy (link at the bottom of that screen, or agentsroom.dev/en/privacy). Note that your agents and tasks stay tied to your account if you use the account/sync features, which is separate from usage analytics.