prompt-guard

What problem does it solve?

This Skill proactively identifies and classifies potentially harmful prompt injection attempts within text data, safeguarding AI systems from malicious inputs.

Core Features & Use Cases

• Prompt Injection Detection: Scans text for known patterns of prompt injection, such as role overrides, command execution attempts, and evasion techniques.
• Risk Classification: Assigns a risk level (dangerous, suspicious, safe) to each detected pattern, allowing for prioritized response.
• Use Case: Before processing user-generated content that will be fed into a sensitive AI model, run it through prompt-guard to flag any attempts to manipulate the AI's behavior or extract confidential information.

Quick Start

Use the prompt-guard skill to scan the provided text for any injection attempts.