The Agent Skills Directory

[PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it ingests untrusted project data.
Ingestion points: The skill processes existing project materials, such as CSS files, Tailwind configurations, and screenshots found within the repository, to extract brand identity information.
Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions when processing these potentially untrusted project files.
Capability inventory: The skill is capable of writing new files to the repository (e.g., DESIGN.md and automated validation scripts) and potentially executing shell commands for prototyping.
Sanitization: There is no mention of sanitizing or validating the content of ingested files before they are interpreted by the agent.
[COMMAND_EXECUTION]: The skill instructs the agent to generate and execute local validation scripts to verify design tokens and accessibility standards.
These scripts are intended to remain in the repository as standing artifacts that execute during future project modifications.
[EXTERNAL_DOWNLOADS]: The skill provides links to several external font foundries and design blogs (e.g., velvetyne.fr, fontshare.com, collletttivo.it) to facilitate the download of typography assets.
While these represent external data sources, they are aligned with the skill's primary purpose of design research and involve specific assets rather than executable code from unknown sources.

develop-design-md