PhoeniksRedTeamers

PhoeniksRedTeamers is an advanced AI red-teaming and prompt-injection analysis platform designed to help researchers, security engineers, and LLM developers test model robustness. The system combines a multi-model chat interface with automated detection of jailbreak attempts using a custom PHOENIX risk-scoring engine. Users can chat with various OpenRouter models while the app evaluates each response for potential prompt-injection patterns, harmful content, or obfuscation-based attacks such as leetspeak, base64, emoji encoding, or multi-layer transformations. The built-in prompt library enables dynamic replacement of payloads for mass testing across hundreds of attack templates. Additional tools include text masking utilities and a real-time analytics dashboard that visualizes risk trends, jailbreak rates, and policy violation patterns using randomized example data. Overall, the platform serves as a practical, educational, and research-oriented environment for understanding LLM vulnerabilities, strengthening AI safety, and experimenting with ethical red-team attack methodologies.