The Agent Skills Directory

[SAFE]: The skill is entirely instruction-based and contains no executable code, Python scripts, or automated tools, as confirmed in the SKILL.md manifest.
[EXTERNAL_DOWNLOADS]: The README.md outlines an installation process using the npx lovstudio command. This facilitates the download and activation of the skill through the vendor's official infrastructure.
[PROMPT_INJECTION]: The skill ingests external design images, which serves as a potential surface for indirect prompt injection. The skill's limited output scope (markdown reports) and lack of executable capabilities mitigate this risk.
Ingestion points: User-provided reference images (SKILL.md, Step 1).
Boundary markers: Not explicitly defined in the prompt instructions.
Capability inventory: Restricted to natural language visual analysis and text generation; no system-level or network capabilities are invoked.
Sanitization: No sanitization of image metadata or pixel data is specified.

lovstudio-visual-clone