The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The skill's documentation instructs users to execute installation and troubleshooting commands with elevated privileges, specifically 'sudo apt install ffmpeg' and 'sudo npx playwright install chromium'.
[EXTERNAL_DOWNLOADS]: The skill fetches design assets and brand guidelines from an external, non-whitelisted domain (getdesign.md). It also pulls core libraries (React, Babel) from the unpkg.com CDN to support browser-based prototyping.
[COMMAND_EXECUTION]: The skill utilizes several shell-based tools for its core functionality, including 'playwright' for generating screenshots and PDF exports, and 'ffmpeg' for video and audio processing. It also manages a cleanup hook that automatically deletes log files in the '.playwright-mcp/' directory.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests untrusted design content from external URLs which could contain malicious instructions designed to influence the agent's behavior during the prototyping phase.
Ingestion points: The 'Progressive Brand Loading' feature fetches markdown files from 'https://getdesign.md/brands/{brand-name}.md' directly into the agent context.
Boundary markers: The skill does not define specific delimiters or instructions to ignore embedded commands within the fetched design files.
Capability inventory: The skill has the capability to write files (exports and design context), read local configurations, and execute shell commands for media processing.
Sanitization: There is no evidence of sanitization or structural validation performed on the design tokens or markdown fetched from the external brand repository.

cc-design-html-prototyping