android-pentest

Fail

Audited by Gen Agent Trust Hub on Feb 28, 2026

Risk Level: HIGHEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill's setup scripts (frida_server_manager.py, build_module.sh, service.sh) download the Frida server binaries directly from the official Frida GitHub repository. It also fetches the Drozer agent APK from the WithSecureLabs GitHub repository. These sources are recognized as well-known security tools and repositories.
  • [COMMAND_EXECUTION]: The workflow orchestrators (base.py, full_assessment.py, etc.) extensively use Python's subprocess module to execute adb shell commands and frida-tools. This is the primary intended behavior of the skill to interact with and instrument Android devices during an audit.
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface (Category 8) because it ingests untrusted data from applications under test. * Ingestion points: data_leakage_scanner.py reads system logs (logcat), while full_assessment.py and crypto_auditor.py dump database contents and preference files. * Boundary markers: The scripts do not consistently use delimiters or 'ignore instructions' warnings when processing this data. * Capability inventory: The skill has significant capabilities including arbitrary shell command execution (adb shell) and dynamic instrumentation (Frida). * Sanitization: There is no evidence of sanitization for data extracted from target applications before it is incorporated into the internal context for report generation.
Recommendations
  • HIGH: Downloads and executes remote code from: https://api.github.com/repos/frida/frida/releases/latest - DO NOT USE without thorough review
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 28, 2026, 12:54 AM