android-mobile-security-sandbox-jamboree

Warn

Audited by Gen Agent Trust Hub on Sep 18, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPRIVILEGE_ESCALATIONPERSISTENCEDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill clones a framework from an unverified third-party repository at 'github.com/hero-mike/Android-Mobile-Security-Sandbox-Testing.git' and downloads Frida server binaries from 'github.com/frida/frida/'.
  • [COMMAND_EXECUTION]: Extensively uses privileged commands to modify the Android system environment, such as mounting the system partition as writable ('mount -o rw,remount /system') and setting global network proxies via ADB.
  • [PRIVILEGE_ESCALATION]: Acquires and uses root permissions ('su') to disable platform security controls, including root detection and SSL certificate validation, and modifies the system certificate store.
  • [PERSISTENCE]: Establishes persistence by installing custom initialization scripts in Magisk service directories ('/data/adb/service.d/') and appending configurations to the system 'build.prop' file.
  • [DYNAMIC_EXECUTION]: Employs Frida and Objection for runtime instrumentation, which involves injecting JavaScript hooks into active application processes to manipulate memory and execution flow.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests data from untrusted sources, including database query results, intercepted HTTP traffic, and hooked method arguments, and processes them through shell commands and scripts without explicit sanitization or boundary markers.
  • [DATA_EXFILTRATION]: Includes specific functionality for extracting sensitive data such as authentication tokens from shared preferences and dumping full process memory to the local filesystem for external retrieval.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Sep 18, 2026, 05:06 PM