ai-safety-redteam-advanced
Installation
SKILL.md
Skill: AI Safety Red Team — Advanced
Supplementary Files:
payloads.md— 60+ ready-to-use LLM attack payloads (prompt injection, jailbreak, RAG poisoning, model extraction)test-cases.md— 12 structured test cases covering OWASP LLM Top 10guides/— Deep-dive guides on advanced techniques
Summary
Advanced AI safety red team operations against Large Language Model (LLM) systems, agent frameworks, and Retrieval-Augmented Generation (RAG) applications.
Tools: garak, PyRIT, promptfoo, Giskard, Counterfit, Lakera Guard, Llama Guard, Neptune, GPT-4 as judge, custom fuzzers
Domain: ai-safety
OWASP: LLM Top 10 (2025)
MITRE ATLAS: Multiple adversarial techniques