The Agent Skills Directory

[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests untrusted user data and interpolates it into a generated HTML artifact.
Ingestion points: User-provided component names, descriptions, and connection details extracted during the workflow described in SKILL.md.
Boundary markers: The skill does not define delimiters or provide 'ignore embedded instructions' warnings for the data being processed.
Capability inventory: The skill generates and writes .html files and suggests the use of Playwright for browser-based rendering in its error handling section.
Sanitization: There are no instructions provided to the agent to escape HTML characters or JavaScript string delimiters when populating the {{DIAGRAM_TITLE}}, {{LEGEND_ITEMS}}, or the connections array in assets/template.html.
[COMMAND_EXECUTION]: The documentation in SKILL.md suggests running playwright install chromium to address rendering failures. This is a legitimate command for a well-known browser automation tool used to manage environment-specific dependencies.

architecture-diagram