Account security
Sign-in methods, active sessions, confirmation codes on high-risk actions, and the security log, all from Settings.
Settings > Security is the account's own control panel: how you sign in, where you are currently signed in, and a record of both.
Sign in
A Xuda account signs in with a password, a one-time code emailed to your address, or a linked Google or GitHub account.
Password
8 to 100 characters, at least one uppercase letter, at least two digits, no spaces. The sign-up and change-password screens check the same rule the server enforces, so a password that clears the checklist on screen is never rejected afterward.
Email code
Enter your address, receive a 6-digit code by email, and type it in. The reply is worded the same way whether or not the address has an account, so the sign-in screen never reveals which emails are registered. A code can be resent after a short cooldown, and sending a new one retires the previous code.
Google or GitHub
Continue with a linked Google or GitHub account instead of typing a password.
The Forgot password link on the sign-in screen emails a reset link. Opening it lands on a Change password screen with the same live checklist as sign-up, and the button stays disabled until every rule is met.
Active sessions
Settings > Security > Sessions lists every place your account is currently signed in: a device icon for desktop, tablet or mobile, a location built from the sign-in (city and country with a flag, or "Unknown location" when neither is known), the IP address, and how long ago it started. A session opened somewhere other than a browser (the CLI, a deployment server) carries a badge saying so, and the session you are reading this on is always labeled This device, pinned to the top of the list.
End one session
Press End session on your own row, or Sign out on any other. Ending your own session signs you out immediately and returns you to sign-in; ending another one closes it without affecting yours.
Sign out everywhere else
Sign out others, at the foot of the list, closes every session except the one you are using in a single confirmed step. It can never close the session you are reading it from.
Confirmation codes on high-risk actions
A short list of irreversible actions asks for a one-time, 4-digit code emailed to the account owner before they run: destroying a project, restoring a backup over the project it came from, and reinstalling a deployment server. Creating a new project from a backup, unlike restoring over one, does not ask for a code. The flow is the same wherever it appears:
Start the action
The first attempt runs with no code and comes back asking for one; nothing has happened yet.
Check your email
A 4-digit code arrives at the account owner's address. There is no expiry, so if it does not arrive, repeating the action sends a new one.
Enter the code
Type the 4 digits into the confirmation dialog and confirm. A wrong code shows an inline error and lets you try again; the action only runs once the code matches.
See Advanced for destroying or reinstalling a project, and Backups for restoring one.
AI provider keys
Also under Settings > Security: your own Anthropic and OpenAI keys, used to run Xuda's AI coding engines on your account rather than Xuda's. A key is write-once, shown afterward only as a Connected badge with its last 4 characters and when it was added, and can be replaced or removed but never read back in full. This is a different credential from the platform API keys covered next: an AI provider key lets Xuda call an AI provider on your behalf, a platform API key lets your own code call Xuda.
Security log
Settings > Security > Sensitive logs is the account's audit trail: sign-ins and authentication events, project updates, and branding changes, grouped by day with a filter bar, a preview pane, CSV export, and a delete action.