Evading an AI SOC with Sable from Vulnetic
Marketing blog post for existing product, not a tool you can actually try or verify.

AI pentesting for Android when web app tools dominate the market.
Mobile security researchers, Android developers
NowSecure · Appknox · Datadog Mobile Security
I have noticed there are growing number of AI native pentesting tools for web apps but very few for mobile or Android. With more mobile apps being shipped quickly due to vibe coding, I wanted to build an AI native security tool specifically for Android apps.
Exfault combines static and dynamic analysis with AI agents using tools like adb,jadx, apktool for static analysis and reverse engineering, frida for dynamic analysis, hermes-dec for React native decompilation. The AI agents have access to real Android emulators to peform navigation, explore functionality and validate vulnerabilities before reporting them improving both the quality of reports and also the rate of false positives.
Instead of uploading an apk or aab, you can simply enter an Android package name (com.example.app), Our backend automatically acquires a compatible build, installs it in an emulator so the agents can test your app.
For authenticated testing, you can provide test credentials and the agent will automatically sign in and continue exploring the authenticated attack surface. I'm also working on a human-in-the-loop login helper for more complex authentication flows involving MFA, Email verification etc.
There's a free demo available if you'd like to try it on your own app.
I'd really appreciate your thoughts and feedback!
Marketing blog post for existing product, not a tool you can actually try or verify.
It actually looks for the weird stuff that trips up LLM agents — invisible Unicode, bidi overrides, embedded curl|bash one-liners, exfil links — and pairs a static skill scanner with a real-time interception flow that forces human approvals. The CLI-first approach (npx safeclaw start) plus Socket.IO alerts and per-command allow/deny decisions show practical thinking about developer workflows; I want to see model/false-positive metrics and enterprise integration docs next.
Autonomous agents doing peer review — nobody's asked what happens if AI reviews AI.
Drops autonomous experimentation into Cursor without installing new frameworks or complex agents.
Kernel-level intent tracking stops AI exfiltration where EDR and Docker fail.
Runtime pentesting on every PR beats SAST — actual exploits, not hypothetical vulnerabilities.