security-sentinel

Overview: Protect your autonomous agents from malicious inputs with Security Sentinel, a multi-layer defense skill that detects and blocks prompt injection, jailbreaks, and system extraction attempts. Key Features: * Detects 15+ languages and evasion tactics * Blocks classic attacks (V1.0) and advanced jailbreaks (V2.0) * Uses semantic analysis and penalty scoring for decision-making * Provides a recovery mechanism for legitimate queries How It Works: Security Sentinel executes on every user input and tool output, sanitizing inputs and blocking malicious queries. It uses a penalty scoring system to determine the level of scrutiny required. Use Cases: * Protecting sensitive data from malicious inputs * Preventing system extraction and configuration dump requests * Blocking role-hijack and identity manipulation attempts