Gautam-R-Patil/S.O.V.A---TRUSCOR ? reverse-engineered prompt

Build me a desktop app and command line tool for autonomous AI security testing.

I want it to let me point at an AI product or web app, have it explore the interface, try different attack style prompts and flows, and then score how resistant it is. It should keep a running session log, capture screenshots or DOM state, and produce a clear report at the end with a risk score, key findings, and a PDF summary I can share.

The app should feel like a dark themed war room with a live terminal, a score gauge, chat style status updates, and basic metrics. It also needs to support running from the command line for automated testing, store session data locally, and work with different AI providers if needed. If you need current docs for any libraries or browser automation pieces, look them up online.