android

Pass

Audited by Gen Agent Trust Hub on Mar 7, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATION
Full Analysis
  • [PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect prompt injection as it processes external Android project files and code. • Ingestion points: Android project configuration and source files (e.g., build.gradle, Java source). • Boundary markers: None specified. • Capability inventory: Shell command execution via adb, gradle, and sdkmanager, plus network operations to Google APIs. • Sanitization: No evidence of validation or sanitization for project content.
  • [EXTERNAL_DOWNLOADS]: Fetches standard dependencies and SDK components from Google's official repositories and package registries using gradle and sdkmanager.
  • [COMMAND_EXECUTION]: Utilizes standard Android development tools (adb, gradle, sdkmanager) for legitimate tasks such as building apps, device management, and package installation.
  • [DATA_EXFILTRATION]: Correctly manages API authentication for the Google Play Developer API by instructing the use of environment variables ($GOOGLE_API_KEY) rather than hardcoding secrets.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 7, 2026, 05:45 PM