Just Ask Curious Code

Audit and defend LLM-powered applications against system prompt extraction attacks using the JustAsk framework's UCB-based probing strategy. Helps security engineers red-team their own AI deployments and harden prompt confidentiality. Trigger phrases: - "Test if my system prompt can be extracted" - "Red-team my LLM application for prompt leakage" - "Audit my chatbot's system prompt security" - "Harden my AI agent against prompt extraction" - "Check if my API endpoint leaks its instructions" - "Build a prompt extraction defense layer"

ndpvt-web Updated 14 repo stars

File contents

ndpvt-web/arxiv-claude-skills commit 6530422121

Frequently asked questions

npx get-skill add ndpvt-web/just-ask-curious-code