shiny-bluetooth
Pass
Audited by Gen Agent Trust Hub on Nov 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides instructions for interacting with external Bluetooth devices, which constitutes a potential surface for indirect prompt injection if data received from a device (such as its name or stream content) contains malicious instructions.
- Ingestion points: Data enters the agent's context through
IBluetoothDevice.Name,IBluetoothDevice.Address, and theBluetoothConnection.Stream(found inSKILL.md). - Boundary markers: The instructions explicitly warn developers that RFCOMM is a byte stream and recommend framing the protocol with line endings or length prefixes to ensure data integrity (found in
SKILL.md). - Capability inventory: The skill allows the agent to generate code for device discovery, pairing, and establishing RFCOMM client/server connections (found in
SKILL.md). - Sanitization: While the skill advocates for protocol framing to handle stream fragmentation, it does not explicitly instruct on sanitizing device metadata (like names) before processing, though this is a standard risk for communication-focused skills.
Audit Metadata