The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the uvx command to execute the yutori-mcp utility, allowing the agent to perform clicks, swipes, and text entry within the iPhone Mirroring window.
[EXTERNAL_DOWNLOADS]: The uvx command fetches the yutori-mcp package from a remote registry at runtime. As this package is a direct vendor resource from 'yutori-ai', it is considered a functional requirement of the skill.
[INDIRECT_PROMPT_INJECTION]: The skill processes visual content from the iPhone screen, which represents an ingestion point for untrusted data that could potentially contain malicious instructions. Ingestion points: Captures and analyzes visual frames of the 'iPhone Mirroring' application window (SKILL.md). Boundary markers: The instructions provide specific interaction constraints and explicitly forbid the agent from exploring apps or data outside the user's requested task. Capability inventory: The skill can simulate user input and navigate the iPhone interface using the yutori-mcp tool. Sanitization: The instructions direct the agent to avoid sensitive app categories (e.g., banking, health, messages) and advise the user to remove private widgets or notifications before use.

yutori-iphone-mirroring