android-ble
Pass
Audited by Gen Agent Trust Hub on Aug 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to process external data from Bluetooth Low Energy (BLE) peripherals, which constitutes a typical surface for indirect prompt injection attacks. 1. Ingestion points: The skill parses BLE advertising bytes (SKILL.md Step 4.5) and handles scan results (references/scanning-advertising.md). 2. Boundary markers: No specific boundary markers are required for raw radio data ingestion. 3. Capability inventory: The skill manages GATT operations (reads/writes) and audits local code via a Python utility. 4. Sanitization: Step 4.5 of the main instructions explicitly requires defensive parsing of advertising bytes and the use of versioned identifiers.
- [SAFE]: The included inspection utility (scripts/inspect-project.py) is a benign static analysis tool designed to identify BLE patterns in Android projects. It does not utilize any network libraries, perform remote code downloads, or access sensitive files such as credentials or SSH keys.
Audit Metadata