smart-glasses
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFE
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references official SDKs and developer tools from well-known technology companies and hardware vendors, including Meta, Google, and Apple. These include links to official GitHub repositories, package registries (NPM, PyPI, CocoaPods), and Maven Central.
- [REMOTE_CODE_EXECUTION]: Instructions are provided for installing official agent skills from trusted organizations such as
metaandandroidto extend development capabilities. These operations target established repositories and are part of the intended developer workflow. - [DATA_EXFILTRATION]: Network operations described in the code snippets, such as using WebSockets or
fetch, are used for legitimate application purposes like transmitting live captions or telemetry between a phone and the glasses. - [INDIRECT_PROMPT_INJECTION]: The skill identifies surfaces for indirect prompt injection, such as processing phone notifications or web content for display on the glasses. The documentation provides engineering guidance on how to handle these data streams, such as filtering text to verified glyph sets.
Audit Metadata