android-app-reverse

Pass

Audited by Gen Agent Trust Hub on Mar 24, 2026

Risk Level: SAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: Uses ADB (Android Debug Bridge) to interact with connected devices, allowing for package installation, file extraction, and the execution of shell commands, including those requiring root privileges for system-level modifications (e.g., certificate installation).\n- [REMOTE_CODE_EXECUTION]: Enables the injection of custom Javascript instrumentation scripts into Android application processes using Frida to observe and modify runtime behavior, such as cryptographic function calls.\n- [PROMPT_INJECTION]: Presents an indirect prompt injection surface because the agent is instructed to read and analyze message output from Frida hooks (get_messages), which could be influenced by a malicious app being analyzed.\n
  • Ingestion points: Frida message stream processed via get_messages in SKILL.md.\n
  • Boundary markers: Lacks explicit delimiters or instructions to ignore commands embedded within the application's output.\n
  • Capability inventory: Includes ADB shell access, file writing, and network proxy management via frida-mcp and adb-mcp.\n
  • Sanitization: No validation or sanitization of the data retrieved from the instrumentation hooks is described.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 24, 2026, 04:31 AM
Security Audit — agent-trust-hub — android-app-reverse