The Agent Skills Directory

[SAFE]: The skill implements a legitimate WebGL UI component. A thorough review of the provided JavaScript and instructions confirms the logic is focused on rendering and user interaction, with no evidence of malicious intent or hidden behaviors.
[EXTERNAL_DOWNLOADS]: The implementation fetches the Three.js library from unpkg.com, a well-known and trusted CDN. Additionally, the instructions suggest cloning optional assets from the author's GitHub repository (github.com/Rheinmir/uiux-asset). These external references are transparent and originate from trusted or vendor-owned sources.
[INDIRECT_PROMPT_INJECTION]: The skill has a data ingestion surface as it processes external content (images and text) for grid tiles.
Ingestion points: Tile content and images defined in SKILL.md and assets/demo.js.
Boundary markers: None explicitly defined for rendered texture content.
Capability inventory: Uses Playwright for verification and git for asset retrieval.
Sanitization: Not applicable for this rendering-only use case as the data is not interpolated into agent prompts.
[COMMAND_EXECUTION]: The workflow includes the use of standard development tools such as git for asset management and Playwright for UI testing. These commands are used appropriately within the context of component development and verification.

infinite-webgl-grid