Automatic Assessments

Automatic Assessments are the unattended mode of the project's AI evaluation. Availability depends on your license, credits, project access, and the agent types enabled for your organization.

Before starting

Review the project's included and excluded scope, environment context, authorized test boundaries, and available credits. Confirm that the selected agent type is appropriate for the engagement. Clients use their own authorized project context; providers use the authorized client engagement.

Configure an assessment

In the project's AI tab, the start form offers agent type, model combination when available, run mode, depth, credit budget, connector when required, and additional instructions. Choose Automatic for an unattended run.

These options shape the evaluation within the project's configured permissions and boundaries. Do not use instructions to expand the authorized scope.

Monitor progress

Use the streaming view to follow stage progress, tool results, recorded observations, artifacts, and candidate findings. A completed run is not a guarantee that every possible vulnerability was tested or that every candidate is valid.

Controls

Automatic mode has stage controls, redo controls, and stop. Chat is available when a run is taken over into the assisted workflow, rather than as an always-on automatic-mode control. Stopping preserves the results already recorded; check the final run status and credit transaction afterward.

Review the outcome

Open the project's findings and review candidates against their evidence. Accept valid findings, reject invalid ones, request additional evidence, or merge duplicates where permitted. Track remediation through the normal issue lifecycle.

See AI-assisted review, Credits, and Run modes.

Recurring assessments

If schedules are available to your organization, use the AI schedules feature to create recurring evaluations. AI schedules are separate from the platform's internal surface-monitoring schedules.