homekit

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill facilitates the processing of data from smart home accessories and characteristics, creating an indirect prompt injection surface where malicious device-provided content could influence agent behavior.
  • Ingestion points: The skill reads external data via HMCharacteristic.value, HMAccessory.name, and HMRoom.name in SKILL.md and references/matter-commissioning.md.
  • Boundary markers: There are no explicit boundary markers or instructions to ignore embedded commands within the characteristic values or accessory metadata.
  • Capability inventory: The skill provides snippets for writing characteristic values (writeValue), creating action sets (addActionSet), and commissioning new devices (MatterAddDeviceRequest.perform).
  • Sanitization: The implementation patterns do not include sanitization, validation, or escaping of the ingested strings from the accessory characteristics.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 01:48 AM
Security Audit — agent-trust-hub — homekit