# JoySafeter

An enterprise AI agent platform for building, orchestrating and running security agents at scale, such as automated APK vulnerability analysis or penetration testing.

JoySafeter is an AI-native platform for building, orchestrating and running security agents, aimed at replacing brittle scripts and single-agent tools with coordinated agent teams for complex security scenarios.

A DeepAgents mode adapts to what an agent finds in real time rather than following a fixed playbook, and more than 200 tools are wired in through the Model Context Protocol without custom glue code. Example workflows include an APK vulnerability detection agent that runs MobSF static analysis, extracts risk signals such as permission abuse and hardcoded secrets, deep-validates high-severity findings with Frida dynamic instrumentation, and generates an OWASP Mobile Top 10 report with no engineer involved; and a penetration testing agent that plans and executes tests against an authorized target and adapts its next step, for example triggering authentication-bypass testing automatically when it finds a login page.

Agents run inside a configured sandbox image, and both cases finish with a downloadable report.

Website: https://github.com/jd-opensource/JoySafeter