Data security
- Row policies built with a condition builder, using system variables for the current user.
- Table and column hiding for sensitive fields.
- Test access: simulate what a user or role will see before you publish a policy.
- Access control lists on content, plus Creator, Reader, and AI user types.
Identity
- Single sign-on with OAuth2, SAML 2, CAS, and LDAP.
- JWT sign-in for embedded users, signed by your backend.
- Personal tokens for MCP clients: per user, shown once, expiring, and individually revocable.
Oversight
- Audit log of administrative and content actions.
- Tenant-wide settings with conflict checks and an audit trail.
- Usage dashboards and quotas for AI questions.
Hardening in 10.00
10.00 evaluates system variables per user, limits custom SQL and new connections for non-admins, requires Delete permission for deletions, and restricts LLM API keys to admins and the AI Agent. We recommend every installation upgrade.
Security and compliance
How we handle data, sub-processors, and our compliance roadmap. Security & compliance →