android-reverse-engineering

Pass

Audited by Gen Agent Trust Hub on Mar 16, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted external data in the form of Android binaries, creating an attack surface for indirect prompt injection. \n
  • Ingestion points: APK, XAPK, JAR, and AAR files are processed and decompiled (SKILL.md). \n
  • Boundary markers: The workflow lacks clear delimiters or instructions to prevent the agent from obeying malicious commands embedded in the decompiled source code or metadata. \n
  • Capability inventory: The agent uses decompilers (jadx, vineflower) to extract logic and document API endpoints (SKILL.md). \n
  • Sanitization: There is no evidence of sanitization or filtering applied to the extracted code strings before they are processed by the agent. \n- [COMMAND_EXECUTION]: The skill relies on external command-line tools to perform its core tasks. \n
  • Evidence: The workflow involves executing jadx, Vineflower, Fernflower, and dex2jar to decompile and analyze artifacts (SKILL.md).
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 16, 2026, 06:30 PM