Hands-on playbook · 55 min
garak Playbook: Scan an LLM Endpoint, Read the Results Honestly and Make It a Regression Test
A hands-on garak deep dive: point it at your own REST endpoint, select probes with the --spec grammar, keep runs fast and repeatable with a YAML config, read the report, hit log and hit@k correctly, triage hits with a script, and turn the scan into a baseline-and-regression check.
- Point garak at your own application's REST endpoint with a generator config
- Select exactly the probes your threat model needs with the --spec grammar
- Write a YAML run config that makes scans fast, repeatable and reviewable
- Read the report, the hit log, Z-scores and hit@k without overclaiming
- Triage hits per probe and detector with a short offline script
- Turn a one-off scan into a baseline and a regression gate for releases
- Know when to write a custom probe or detector, and what each must contain
- 01Under the hood
- 02Hands-on lab
- 03Choosing probes from a threat model
- 04Recipes
- 05Rollout runbook
- 06Troubleshooting
- 07Interview drill
- 08Sources
Every worked answer in all 17 topics, every lesson in every course, the full concept curriculum, hands-on tool playbooks, the complete practice-test bank, and the downloadable security field guides.
6 months · One payment · No auto-renewal
An independent playbook for interview preparation and learning. AISecurityInterviews is not affiliated with or endorsed by NVIDIA (open source, Apache-2.0), and product names are trademarks of their owners. Use these procedures only on systems you own or are authorised to test, and check the vendor's documentation before changing production.