Test AI and LLM systems for offensive security risks like prompt injection.
npx skills add https://github.com/noname300989/Security-Claw --skill ai-offensive
Please help me install this Agent Skill. Skill: ai-offensive Source: https://github.com/noname300989/Security-Claw/tree/main/skills/ai-offensive Command: npx skills add https://github.com/noname300989/Security-Claw --skill ai-offensive
This Skill addresses the critical security vulnerabilities inherent in AI and LLM systems, protecting against sophisticated attacks that can compromise data, functionality, and user trust.
Use the ai-offensive skill to test the AI chatbot at https://example.com/chat for prompt injection vulnerabilities.
High-intent search queries and answers about installing and using this skill.
To test an LLM chatbot for prompt injection vulnerabilities, you can use this Skill to simulate attacks designed to override AI instructions and hijack agent behavior across chat APIs. It validates security against malicious instruction overrides before deployment.
RAG poisoning involves malicious data corrupting Retrieval-Augmented Generation pipelines. This Skill assesses RAG security by testing if corrupted retrieval data can manipulate your AI system's generated outputs and compromise pipeline integrity.
Yes, you can run red teaming tests for excessive agency in AI agents. This Skill tests whether AI agents can be tricked into performing unauthorized actions, covering vulnerabilities listed in the OWASP LLM Top 10 and Agentic AI Top 10.
Yes, this Skill covers the OWASP LLM Top 10, Agentic AI Top 10, and MCP Top 10 vulnerabilities. It tests chat APIs, agent communication, and RAG pipelines to validate security against malicious instruction overrides and unauthorized actions.
To check your AI system for system prompt leakage, this Skill identifies weaknesses that allow extraction of sensitive system prompts. It simulates attacks across chat APIs to ensure your proprietary instructions cannot be extracted by malicious users.
jinyimeng01Test AI and LLM applications for adversarial security weaknesses.
vanillacake369Identify security vulnerabilities in codebases with prioritized remediation guidance.
ricardorocheDetect prompt injection attempts in AI and LLM applications.
hoanghn61Scan source code and dependencies for vulnerabilities, secrets, and malicious code.
