domains / families / security / ai-red-teaming

ai-red-teaming

combined

Run structured AI red team exercises for jailbreak resistance, data exfiltration risk, harmful output controls, and agent tool abuse resilience. Run structured AI red team exercises for jailbreak resistance, data exfiltration

When to use

• Launching a new LLM-powered feature or product
• Evaluating a third-party model before adoption
• Running periodic security assessments of existing AI systems
• Responding to a reported jailbreak or prompt injection incident
• Preparing for compliance audits requiring adversarial testing evidence

When not to use

• Apply guidance only within authorized scope; test destructive steps in non-production first.
• Docs-only import: upstream scripts and templates not bundled.

Skill page

Conflict

The two copies share the same core procedure (threat scenarios, test categories, adversarial prompt library, and a Python harness), but the second copy adds material constraints the first lacks: an explicit authorized-use warning, a mandatory confirmation gate before any probing or data extraction, and a caveat that the harness is a template that must be adapted to a synthetic sandbox with a validated target adapter and independent rubric. The first copy lacks these safety and scope controls. Crowning the first copy would hide a real alternative that adds required pre-flight authorization and sandboxing steps, so the copies should be combined.