Overview
STARS (Smart Threat AI Reporting Scanner) is an AI agent designed to perform vulnerability tests on Large Language Models (LLMs) from SAP AI Core, local deployments, or models from HuggingFace. Its primary objective is to identify and rectify potential security vulnerabilities through automated testing.
Key Features
- Vulnerability Testing: Conducts thorough vulnerability assessments on LLMs.
- Multi-Source Compatibility: Supports AI models from SAP AI Core, local deployments, and HuggingFace.
- User-Friendly Interface: Accessible via a chat frontend or command-line interface (CLI).
- Community-Driven: Open to contributions, feature requests, and bug reports through GitHub issues.
Benefits
- Enhanced Security: Helps organizations identify and mitigate security vulnerabilities in their AI models.
- Flexible Usage: Can be utilized in various environments, ensuring adaptability.
- Open Source: Encourages community involvement and transparency in security practices.
Highlights
- Includes both backend and frontend components for running the agent.
- Comprehensive documentation available to assist users in getting started and making the most out of the tool.