Mobile Application Security Testing

Pass

Audited by Gen Agent Trust Hub on Nov 6, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPRIVILEGE_ESCALATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill downloads several well-known security tools and binaries from public repositories.
  • Fetches the Jadx decompiler and Frida server from GitHub releases.
  • Pulls official Docker images for the Mobile Security Framework (MobSF) and Drozer.
  • References a community-provided SSL pinning bypass script via Frida's official codeshare platform.
  • [COMMAND_EXECUTION]: Provides a variety of shell commands to facilitate the security assessment process.
  • Uses adb for device communication, file transfer, and shell access.
  • Executes apktool for decompiling and rebuilding application packages.
  • Employs frida and objection for runtime instrumentation and memory manipulation.
  • [PRIVILEGE_ESCALATION]: Requests elevated permissions for environment configuration.
  • Utilizes sudo to install system packages like adb and apktool on the host machine.
  • Employs chmod +x to set executable permissions on the Frida server binary pushed to the test device.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data (APK files).
  • Ingestion points: Target APK files are ingested during reconnaissance and analysis phases (SKILL.md, Phases 2-10).
  • Boundary markers: None provided for the content of the analyzed files.
  • Capability inventory: The skill enables full device shell access via ADB, runtime code injection via Frida, and local file system modification via Apktool (SKILL.md, Phases 1, 6, 8, 10).
  • Sanitization: No explicit sanitization or validation logic is defined for the processed APK data.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 6, 2026, 03:30 PM