Find security weaknesses in your AI systems.
What it does
Guides security assessments of AI and machine-learning systems to identify model, supply-chain, privacy, and agent-security weaknesses.
When to use it
Use when assessing model supply-chain attacks, adversarial examples, poisoning, model theft, data privacy attacks, or autonomous-agent security risks.
How to use it
It changes how the agent assesses an AI or machine-learning project by directing checks according to its model pipeline, deployment, training data, and interfaces.
What you provide
- An existing project
Latest review
What you need · 1
Directs the agent to load this skill when assessing an LLM or chatbot for prompt injection, jailbreaking, and tool abuse; it is not required for non-LLM assessments.
Access · 0
No additional actions listed.