In simple terms
This guide explains garak Auto-populate in clear language. Normalize authorized garak probe results into an AI Application Security project. Follow the steps in order, then use the final checklist to confirm that everything worked.
Why this matters
Use this page when you want a clear answer, a practical sequence, and a way to check your result. You do not need to learn every voiqq feature before you begin.
What this solves
AI security probe output may include sensitive prompts, variable responses, detector signals, and model behavior that does not automatically represent an exploitable application vulnerability.
How voiqq helps
AI Security Auto-populate organizes supported garak results into ordinary voiqq candidates, retains minimum necessary prompt and response evidence, and suggests AISVS mappings for human review.
Before you begin
- Authorization to test the model, agent, API, or integration
- A configured AI Security project and intended AISVS Library
- A supported garak result or safe configured test target
- A plan to remove secrets and unnecessary personal data
Step-by-step
- Open Auto-populate in the AI Security project
- Choose the supported report or configured test flow
- Review probe, detector, target, and response context
- Confirm reproducibility and application impact
- Confirm or adjust the AISVS mapping
- Accept, assign, remediate, and retest legitimate findings
Check your result
- Sensitive prompts are minimized
- Model behavior is distinguished from integration behavior
- Accepted findings can be reproduced
- Severity reflects practical impact rather than probe success alone