apk-reverse

Pass

Audited by Gen Agent Trust Hub on Oct 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from Android APK files, creating a surface for indirect prompt injection where the AI might encounter and follow instructions embedded in the analyzed code or logs. Findings include:
  • Ingestion points: Extraction and reading of Java source code (via jadx), XML manifest files (via apktool), and system logs (via adb logcat) from third-party APKs.
  • Boundary markers: No specific delimiters or instructional warnings are present to help the agent distinguish its primary instructions from the content of the analyzed files.
  • Capability inventory: The skill has access to powerful tools including 'adb', 'frida', and 'apktool', which can perform file system modifications, command execution, and network operations.
  • Sanitization: There is no evidence of sanitization or filtering of the data extracted from APKs before it is presented to the agent.
  • [EXTERNAL_DOWNLOADS]: The skill includes an 'On-Demand Bootstrap' feature that fetches software from remote servers to set up the environment. Evidence: The skill description states it will automatically download and install 'jadx' and 'apktool' from GitHub releases and 'frida' via the Python package manager (pip).
  • [DYNAMIC_EXECUTION]: The skill employs a dynamic bootstrap mechanism that installs dependencies at runtime if they are missing. Evidence: The 'On-Demand Bootstrap' section describes the automated setup of the toolchain through scripts that download and execute installers and package managers.
  • [COMMAND_EXECUTION]: The skill's primary operations are centered around the execution of various command-line interface (CLI) tools. Evidence: Extensive use of 'adb' for device management, 'frida' for dynamic hooking, and 'pwsh' (PowerShell) for automation scripts throughout the workflow.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 14, 2026, 03:07 PM