android-walkthrough-video
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No security vulnerabilities were detected. The skill performs legitimate Android automation tasks and follows security best practices for the tools it utilizes.\n- [COMMAND_EXECUTION]: The skill executes local utilities including
adb,ffmpeg, andffprobe. These commands are constructed safely by passing arguments as lists to thesubprocessmodule, avoiding shell-injection risks. Commands executed on the Android device viaadb shellare properly escaped usingshlex.quote.\n- [INDIRECT_PROMPT_INJECTION]: The skill has a data ingestion surface through caption text provided by the agent. While explicit boundary markers are absent, the capability inventory (includingadbandffmpeg) is not exposed to this untrusted data; the text is exclusively processed by the Pillow library for image rendering and is never used in shell commands or sensitive tool parameters.
Audit Metadata