ai-security
Installation
SKILL.md
Skill: AI/LLM Security
Supplementary Files:
payloads.md— Prompt injection templates, jailbreak frameworks, model extraction probes, RAG poisoning documents, and garak/promptfoo commands organized by attack categorytest-cases.md— Structured test cases covering direct injection, indirect injection, jailbreaking, system prompt extraction, RAG poisoning, and full AI application auditguides/— Deep-dive methodology guides for systematic AI/LLM security assessmentExtended Guides (
guides/):
guides/llm-attack-methodology.md— Attack surface identification, threat modeling, attack chain construction, jailbreak taxonomy, evidence collection, defensive bypass detection, and AI vulnerability reportingguides/ai-security-jailbreak-research.md— LLM jailbreak research methodology: prompt injection taxonomies, adversarial suffix attacks (GCG/AutoDAN/PAIR/TAP), multi-turn Crescendo patterns, multimodal jailbreaks, real-world incidents, defense tooling, and eval harnesses (HarmBench/AdvBench/StrongREJECT)
Summary
Ai Security skill domain covering ai operations.
Tools: garak, promptfoo, llm-guard, picklescan, ffuf, burpsuite, Custom Python, Burp Suite