macos-tv
Pass
Audited by Gen Agent Trust Hub on Mar 13, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill executes
osascriptcommands to interface with the macOS TV.app. These operations are limited to the app's automation dictionary for playback control (play, pause, seek) and library management (list movies, search tracks), which is the intended and safe use for this skill. - [PROMPT_INJECTION]: An indirect prompt injection surface is present as the skill ingests metadata (e.g., movie titles and artist names) from the local TV library and returns it to the agent context. While this data is untrusted, the skill's capabilities are restricted to media control. 1. Ingestion points:
scripts/library/movies.applescript,scripts/library/tv-shows.applescript,scripts/track/current.applescript. 2. Boundary markers: Absent in the script output. 3. Capability inventory: Localosascriptexecution for TV.app control. 4. Sanitization: None performed on the retrieved metadata.
Audit Metadata