Security
Last updated: 20 November 2026
1. Accounts and sign-in
1.1 People sign in with a six-digit code sent to their email address. Rondai does not use passwords. Codes expire after 10 minutes, allow five attempts and work once.
1.2 Session tokens are stored only as hashes. Sign-in codes and invitation links are checked against hashes; while an email carrying one is waiting to be sent, its content is held encrypted and is cleared once the email has been delivered.
1.3 Sign-in responses never reveal whether an email address has an account.
1.4 Repeated sign-in attempts and code requests are limited per address and per network.
1.5 Sessions end after 7 days without activity and after 30 days at most. You can see your active sessions and end any of them.
2. Organisations and roles
2.1 Every workspace record belongs to one organisation, and every request is checked against the organisation of the person making it. People see only the organisations they belong to. Records that belong to no organisation, such as demo requests and security events, are never shown in any workspace.
2.2 Roles (owner, admin and member) control who can invite people, change roles and manage settings. Only owners can grant the owner role, and an organisation always keeps at least one owner.
3. Records and evidence
3.1 Every change to an organisation's records is written to an audit trail. Rondai only adds to this trail; it never edits or deletes an entry.
3.2 Scan evidence is stored with a SHA-256 digest so it can be checked later. Downloads use short-lived signed links.
3.3 Rondai drafts work on redacted evidence, and nothing a draft says becomes a record until a named person accepts it.
3.4 When Rondai prepares a draft or an answer, the model provider receives only record metadata and the question asked, never cookie values, stored browser values or form entries. Questions and their cited records are kept in the organisation's own workspace, and the audit trail notes that a question was asked without copying its text.
4. Scanning
4.1 Scans run in an isolated browser that visits only the public website being scanned. Requests to private and internal network addresses are blocked.
5. The website and workspace
5.1 Connections use HTTPS. The browser never receives an API key, and sign-in cookies cannot be read by scripts on the page.
5.2 Pages are served with security headers that restrict framing, scripts and browser features.
6. Reporting a problem
6.1 If you find a security issue, email labs@invalcre.com with "Security" in the subject. Please give us reasonable time to fix it before telling others. We will confirm receipt within two working days.