Llm Testing

If you're doing AI safety work or red teaming LLMs, this gives you structured test prompts across bias detection (gender, race, nationality), data leakage, memory recall, and alignment attacks. The wordlists use placeholders you swap out to generate test cases systematically. It's honestly more of a curated prompt collection than a skill that does something, but that's fine because it saves you from writing these from scratch. The real value is having organized categories for security audits and the explicit reminder about authorization, since running adversarial prompts against models you don't own is a fast way to violate ToS. Useful for building internal testing pipelines or academic research, less so if you need automated analysis of results.

Install to Claude Code

npx -y skills add eyadkelleh/awesome-skills-security --skill llm-testing --agent claude-code