BSides talks featuring Prompt Hardener

Open-source toolkit that automatically evaluates, hardens, and adversarially tests LLM system prompts using techniques like spotlighting, random sequence enclosure, instruction defense, and role consistency