Documentation

Security

Ways to sign in, API security, how data is protected, the code check before go-live and security best practices on Nexugen.

Ways to sign in

  • Google and GitHub sign-in through OAuth. Nexugen never sees your password for those accounts.
  • Magic links by email, valid for 30 minutes, single use.
  • Passwords, at least 8 characters, stored scrambled so nobody can read them. A new password account cannot sign in until the address is verified by clicking the emailed link. Reset links are emailed to the address on the account.

Sessions are signed tokens that last 30 days and refresh daily while you use the product. You can add two-step sign-in (a code from an authenticator app or by email) in Settings, Sign-in and security.

API security

  • API keys are shown once and stored as SHA-256 hashes; the platform cannot display a key again. Revoke and recreate rather than share one.
  • A key is read or read and write. Read keys cannot start builds or edits, deploy, change domains or environment, or read environment values.
  • A key acts as its owner and never more: project access, team roles, credits and per-action limits apply exactly as in the dashboard.
  • Every key is rate limited per minute, and usage (call count, last used) is shown in Settings so an unexpected pattern is visible.
  • Webhook deliveries are signed with HMAC-SHA256 over the timestamp and the raw body; verify the signature before trusting a delivery. Secrets are shown once.

The full contract is in the API reference.

Data protection

  • Isolation. Each hosted project runs as its own process with its own database: an isolated Neon project, or a dedicated database on the platform's Postgres. Projects do not share a database or a process.
  • Storage. Project files live in the platform database and in object storage (Cloudflare R2); downloads use short-lived signed URLs.
  • Secrets. Environment variables are stored per project, masked in the dashboard when marked secret, and written to the app only at deploy time. Version snapshots can carry them, which is why snapshots are team-only.
  • Transport. The dashboard, the API, every project subdomain and every verified custom domain are served over HTTPS.
  • Backups. Platform and project databases are backed up nightly.
  • Retention. Session event logs (the live run transcripts) are kept for 30 days. Agent memory is yours to edit or retire at any time. Deleting a project stops its hosting and removes its files and versions.
  • Visibility. Public projects expose their result only. Team material (versions, the thread, proposals, session detail) stays with the team.

Code check

Code check (in the sidebar) reads your app's code and lists weak spots that could let someone in, with a safety score, grouped by kind and kept per project. It uses credits from your balance, more for bigger apps. It only reads; fixing something is an edit you choose to make.

Best practices

  • Create read-only API keys unless a script genuinely needs to write. One key per integration, named for it.
  • Keep keys, tokens and passwords in environment variables marked secret, never in code or prompts.
  • Give stakeholders the Viewer role; give Member and Admin only to people who should change or operate the project.
  • On shared projects, make risky edits as proposals so a second person sees the change summary before it lands.
  • Take a version before a restore, a large edit, or an environment change that touches a database.
  • Make a project Public only when the result is what you want the world to see; everything else stays team-only regardless.
  • Verify webhook signatures and reject deliveries whose timestamp is stale.
  • Run the security scanner before a launch and after any edit that touches auth, payments or file handling.
NextAPI reference