android-bluetooth-expert

Pass

Audited by Gen Agent Trust Hub on Jun 23, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: No patterns of instruction overrides, safety filter bypasses, or jailbreak attempts were found. The instructions are focused strictly on technical guidance for Bluetooth development.
  • [DATA_EXFILTRATION]: No sensitive file paths, credential harvesting, or unauthorized network operations were detected. The code snippets provided are standard Android Bluetooth operations using local APIs.
  • [DYNAMIC_EXECUTION]: The skill mentions the use of reflection (e.g., device.javaClass.getMethod("createRfcommSocket", ...).invoke(...)) and a GATT cache refresh workaround. These are documented techniques used in Android development to address specific hardware bugs and API limitations, rather than for executing arbitrary or malicious code.
  • [EXTERNAL_DOWNLOADS]: The skill references official Android documentation (developer.android.com) and official Google-maintained samples on GitHub (github.com/android/connectivity-samples). These are trusted and well-known sources for development resources.
  • [COMMAND_EXECUTION]: The skill provides instructions for the user to execute standard Android Debug Bridge (ADB) commands for log collection and debugging (e.g., adb pull, adb logcat). These are common development workflows and are not executed by the agent automatically.
  • [INDIRECT_PROMPT_INJECTION]: The skill acts as a code assistant and does not automatically ingest or process untrusted external data (such as scraping arbitrary websites) that could lead to indirect injection attacks.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 23, 2026, 08:33 PM
Security Audit — agent-trust-hub — android-bluetooth-expert