The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill's installation process fetches components and updates from the vendor's GitHub repository (skydashnet/material-design-3-ui-skill). This is a legitimate operation for deploying the skill package and its associated reference documentation.
[REMOTE_CODE_EXECUTION]: The README.md provides installation instructions that involve downloading and executing shell scripts (install.sh and install.ps1) from the repository. These scripts are used to distribute the skill package to various supported AI agent directories.
[COMMAND_EXECUTION]: The skill's documentation includes commands meant to be run by the user to install or uninstall the skill package. These commands involve standard shell piping (curl | bash and irm | iex) to facilitate the setup of the skill's environment.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-supplied design goals and information architecture to generate UI recommendations. While it handles external input, it lacks dangerous runtime capabilities that could be exploited via injection.
Ingestion points: User-provided design prompts and project requirements processed in SKILL.md.
Boundary markers: None present.
Capability inventory: The skill does not have tools or instructions for writing files, performing network operations, or executing arbitrary code during its runtime design tasks.
Sanitization: No sanitization is implemented, but the absence of risky capabilities mitigates the threat surface.
[SAFE]: The core logic and instructional content within SKILL.md and the references/ directory adhere to official Material Design 3 guidelines. There are no signs of credential harvesting, exfiltration, or obfuscated malicious intent.

material-design-3-ui