flutter-device-smoke-test

Pass

Audited by Gen Agent Trust Hub on Aug 14, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes the adb (Android Debug Bridge) command-line tool to interact with connected Android devices. It executes commands to install local APK files, launch specific app activities using the Activity Manager (am start), and capture the device screen (screencap) for verification.
  • [PRIVILEGE_ESCALATION]: The instructions include the use of run-as <packageId>, a standard Android development command that allows the user to assume the identity of a specific debuggable application to inspect its private data directory (app_flutter/). This is used for verifying file creation and storage within the scope of the app's own permissions.
  • [INDIRECT_PROMPT_INJECTION]: The skill defines a process for ingesting external data from the device via adb logcat.
  • Ingestion points: External log data enters the agent context through the logcat command output.
  • Boundary markers: None are present to distinguish log content from instructions.
  • Capability inventory: The skill has access to shell command execution (adb) and local file system writes (saving screenshots to /tmp/app.png).
  • Sanitization: The data is filtered using grep for specific error keywords, but no further sanitization is applied to the log content before analysis.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 14, 2026, 01:42 PM
Security Audit — agent-trust-hub — flutter-device-smoke-test