Skill UI

NVIDIA’s NeMo Guardrails adds runtime safety rails to LLM apps, covering jailbreak detection, hallucination and fact-checking, PII filtering, and toxicity checks via Colang 2.0 rules and integrations for production use on T4 GPUs.

Meta’s Prompt Guard is an 86M-parameter classifier that spots prompt injections and jailbreak attempts before they reach LLMs, filtering user input, third-party data, or RAG documents with multilingual support and sub-2ms latency for fast deployments.

A powerful toolkit utilizing Objection and Frida for authorized mobile penetration testing. It enables deep runtime analysis of iOS applications, allowing users to inspect sensitive data in the keychain, filesystem, and memory. Key capabilities include bypassing client-side protections such as SSL pinning and jailbreak detection without requiring source code access.

Assess AI/ML systems and LLM agents for prompt injection, jailbreaks, model inversion, data poisoning, and tool abuse; map findings to MITRE ATLAS and score robustness.

This tool provides comprehensive AI security scanning using the Tencent AI-Infra-Guard framework. It systematically audits AI infrastructure, scans tools and skills, checks agent security, and performs advanced LLM jailbreak evaluations to identify vulnerabilities and ensure the robustness and reliability of AI systems.

A comprehensive offensive checklist for assessing the security and robustness of AI and Large Language Model (LLM) applications. It covers advanced adversarial techniques such as prompt injection, jailbreaking, model extraction, data poisoning, and analyzing system vulnerabilities across components. Essential for red-teaming and security assessment.

An automated framework for continuous, repeatable red-teaming of LLM applications within CI/CD pipelines. It integrates Promptfoo and DeepTeam to test against recognized vulnerabilities like OWASP LLM Top 10, OWASP Agentic, and MITRE ATLAS (e.g., Prompt Injection, Jailbreaks). This ensures that security vulnerabilities do not regress after code or model changes, providing a critical merge-blocking security gate for production LLM apps.

Implements comprehensive runtime safety defenses for Large Language Model (LLM), RAG, and agent applications. This skill integrates industry-leading guardrail systems—including Llama Guard 3, NeMo Guardrails, and LLM Guard—to inspect and constrain all input and output data. It is essential for production environments requiring protection against adversarial attacks such as jailbreaks, prompt injection (OWASP LLM01), toxic content, and sensitive data leakage.

NVIDIA garak is an open-source LLM vulnerability scanner designed for comprehensive red-teaming and security assessment. It systematically probes large language models (via OpenAI-compatible APIs, Hugging Face, etc.) using thousands of adversarial prompts. It covers critical vulnerabilities such as prompt injection, jailbreaking, data leakage, and toxicity. Use this tool to establish a repeatable security baseline before or after deployment, validate the effectiveness of safety guardrails, or generate quantitative evidence for AI risk assessments.

Detailed methods for red-teaming and penetration testing of LLM applications. This skill extracts sensitive system prompts, secrets, and embedded logic using advanced techniques like jailbreaking, encoding tricks, and automated scanners (Garak, Promptfoo), ensuring compliance with OWASP LLM07:2025.

This tool performs comprehensive safety scanning on untrusted inputs, such as user submissions, API payloads, or webhook data. It detects sophisticated threats like prompt injection, jailbreak attempts, and adversarial attacks using advanced AI defense mechanisms, ensuring data integrity and security before model processing or execution.

A comprehensive methodology for deep security testing of both Android (APK) and iOS (IPA) applications. This guide covers the entire workflow, from initial information gathering and static analysis (decompilation, code structure analysis) to advanced dynamic instrumentation, network interception, and bypassing modern security measures like SSL Pinning, Root/Jailbreak detection, and anti-debugging.