shiny-bluetoothle

Pass

Audited by Gen Agent Trust Hub on Sep 10, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill enables an AI agent to interact with external BLE devices, creating an attack surface where untrusted data from these devices can influence the agent's behavior.
  • Ingestion points: The agent ingests external data through GATT characteristic reads (e.g., ReadCharacteristicAsync), notifications (NotifyCharacteristic), and L2CAP channel data streams (DataReceived). It also processes file metadata such as FileName during L2CAP file transfers.
  • Boundary markers: The skill instructions do not specify or require the use of boundary markers to encapsulate and isolate data received from BLE peripherals.
  • Capability inventory: The skill provides significant capabilities including writing to the local file system (DownloadFile, AcceptDownload) and transmitting data or files back to peripherals (UploadFile, WriteCharacteristic, SendFile).
  • Sanitization: The documentation in reference/api-reference.md explicitly warns that the FileName property in an L2CapFileRequest is untrusted and must be validated before being used as a file path, showing an awareness of potential path traversal or injection risks.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 10, 2026, 11:41 PM
Security Audit — agent-trust-hub — shiny-bluetoothle