The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection by processing external data.
Ingestion points: The skill ingests data from external Website URLs (HTML and CSS) and Figma metadata (metadata, design context, variable definitions) as described in the 'Capture the design' section of SKILL.md.
Boundary markers: The instructions do not define boundary markers (such as XML tags or delimiters) or provide specific instructions for the agent to disregard instructions embedded within the design files or source code being analyzed.
Capability inventory: The skill utilizes shell commands via Playwright, specialized MCP tools for Figma and Pen.dev, and generates a structured DESIGN.md report.
Sanitization: There is no evidence of sanitization, filtering, or escaping of the fetched external content before it is processed by the agent.
[COMMAND_EXECUTION]: The skill instructs the agent to use the Playwright CLI to interact with and capture data from external websites.
Evidence: The instruction npx playwright screenshot --full-page --viewport-size=1440,900 <url> desktop.png in SKILL.md allows the execution of a browser automation tool on arbitrary user-provided URLs.

ryux-analyze