gps-compliance
Pass
Audited by Gen Agent Trust Hub on Jul 28, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructions include shell command templates that interpolate the
$PKGvariable and user-provided data safety payloads (e.g.,gps --package $PKG --confirm call gps_set_data_safety ...). Direct interpolation of user-controlled variables into shell strings can lead to command injection if the input contains shell metacharacters like semicolons or backticks.\n- [PROMPT_INJECTION]: The skill processes untrusted user input such as package names and configuration bodies and passes them to sensitive operations in thegpstool, representing an indirect prompt injection surface.\n - Ingestion points: Package names ($PKG) and JSON safety/recovery payloads are ingested from user input in SKILL.md.\n
- Boundary markers: No specific delimiters or safety instructions are provided to the agent to treat user data as non-executable text.\n
- Capability inventory: The skill can modify data safety labels and manage app recovery actions, which are sensitive operations within the Google Play ecosystem.\n
- Sanitization: The use of shell interpolation patterns like '"$PKG"' is insufficient to prevent all forms of shell-breaking characters or injection attacks.
Audit Metadata