mobile-app-instrumentation

Fail

Audited by Gen Agent Trust Hub on Sep 19, 2026

Risk Level: CRITICALREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPRIVILEGE_ESCALATIONPERSISTENCEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill fetches Frida binaries and security tools from the official Frida GitHub repository and community repositories (e.g., Chariz, opa334, repo.co.kr). It also references the HackTricks security wiki, a common documentation resource in the field.
  • [REMOTE_CODE_EXECUTION]: Instructions are provided to download and execute the Frida instrumentation server as root on connected mobile devices via adb or ssh, which is necessary for runtime analysis.
  • [COMMAND_EXECUTION]: The skill utilizes common security utilities like adb, ssh, apktool, jadx, and Clutch for analyzing binary files and interacting with mobile operating systems.
  • [PRIVILEGE_ESCALATION]: The skill requires and leverages administrative privileges (su, adb root, jailbreak) on target mobile devices to perform memory hooking and access sensitive system components like the Keychain or Keystore.
  • [PERSISTENCE]: The documentation suggests using launchctl on iOS to ensure the Frida server remains active across device reboots.
  • [DYNAMIC_EXECUTION]: Runtime manipulation is achieved through Frida's JavaScript engine, which injects and executes code into live process memory to bypass security gates and trace function calls.
  • [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection. Ingestion points: App code decompilation via jadx and runtime memory hooking via Frida. Boundary markers: Absent. Capability inventory: Shell access (Bash), file modification (Edit, Write), and root command execution on target devices. Sanitization: Absent. Severity: LOW.
Recommendations
  • Contains 2 malicious URL(s) - DO NOT USE
Audit Metadata
Risk Level
CRITICAL
Analyzed
Sep 19, 2026, 04:36 PM