Why I'm building this
I'm Stanislav Miziyev, a security engineer. I've spent my career on the defensive side — threat research, detection engineering, taking apart malicious code and writing the detections that catch it. When AI agents started calling external tools through the Model Context Protocol and loading third-party skills, I saw the same pattern I'd seen before: a powerful new capability shipping with its supply chain wide open.
An agent follows the instructions it reads. If a tool's description tells it to leak a key, it will. The enterprise platforms raising nine figures are solving this for large security teams through sales calls. The developer shipping an MCP server and the small business running a chatbot have nothing. PALANIT is for them — open-source at the core, honest about severity, and built to be trusted because you can read every rule.