debugging

Pass

Audited by Gen Agent Trust Hub on Jul 6, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses standard Android development commands such as ./gradlew and adb for project builds, test execution, and device diagnostics. These commands are consistent with the skill's stated purpose of Android debugging.
  • [EXTERNAL_DOWNLOADS]: The skill mentions an optional companion skill, skydoves/android-testing-skills, which is used for advanced testing guidance. This reference follows documented patterns for modular skill usage and does not involve automated or hidden remote code execution.
  • [PROMPT_INJECTION]: The skill is designed to process untrusted data sources including logcat output, Gradle logs, and Crashlytics reports. While this creates a potential surface for indirect instructions, the skill defines a rigid diagnostic workflow focused on evidence triage rather than executing arbitrary instructions from the log data.
  • Ingestion points: Logcat output, Crashlytics reports, Gradle logs (SKILL.md)
  • Boundary markers: None explicitly specified in the instructions for isolating log content
  • Capability inventory: Shell command execution via ./gradlew and adb (SKILL.md)
  • Sanitization: Standard debugging practices are described without specific filtering rules for log data
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 6, 2026, 12:30 PM
Security Audit — agent-trust-hub — debugging