The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill acts as a bridge between user requests and a diagram generation tool, creating a surface for potential indirect prompt injection through processed data.\n
Ingestion points: The skill ingests untrusted user descriptions and specifications for diagrams as defined in the main SKILL.md instructions.\n
Boundary markers: While the skill provides clear syntax constraints for the generated Mermaid code (e.g., no emojis, no HTML), it lacks explicit boundary markers or instructions to isolate potentially malicious commands embedded in user-provided content.\n
Capability inventory: The skill leverages powerful capabilities including generate_diagram for rendering, use_figma for modifying board content, and get_figjam for reading board data in references/workflow.md.\n
Sanitization: The instructions recommend sanitizing generated output by wrapping special characters in quotes and avoiding specific Mermaid features that could cause issues, providing a degree of basic input management.\n- [SAFE]: The skill is authored by the platform vendor (Figma) and all external resources, such as FigJam board URLs, target the vendor's official domain. The instructions are focused on functional tool usage and include detailed error avoidance guidelines.

figma-generate-diagram