Novo Agents Security

Last updated: July 10, 2026

Service boundary

Customers interact with Novo Agents through an HTTPS API, the server-side SDK, and the console. The agent runtime operates inside Novo-managed infrastructure behind that boundary.

Tenant isolation

Service data is scoped to a workspace, and access is enforced at the data layer and on every route. API keys are scoped to a single workspace.

Encryption and secrets

Traffic is encrypted in transit and stored data is encrypted at rest. Credentials you provide are additionally encrypted at the application layer, decrypted only at the point of use, and never returned by the API. Model-provider credentials remain inside Novo-managed infrastructure and are never sent to customer-configured environments.

Runtime isolation

Agent file and shell execution happens only in workspaces you enable, inside isolated execution environments — Novo-managed sandboxes or environments you host. Requests to environments you host are signed so you can verify their origin.

AI data handling

AI processing is restricted to the disclosed providers in the Novo Agents subprocessors list. A strict data-handling option further restricts processing to configurations verified for zero data retention and no training use, as described in the product documentation.

Webhooks and event delivery

Deliveries to endpoints you configure are signed so you can verify they came from Novo.

Responsible disclosure

Report suspected vulnerabilities to security@novoindustries.co.