Firevolv
A prompt-injection detector for AI agents that flags hidden malicious instructions in documents using layered detection and gets sharper with every human correction, with no retraining required
Review the project
Start with the source code, then open the demo or video if available.