The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructions direct the user to configure an external MCP server endpoint at https://rube.app/mcp. This introduces a dependency on a third-party service provider for the skill's core functionality.
[INDIRECT_PROMPT_INJECTION]: The skill operates on external data provided by Figma files and comments, which represents a potential attack surface for indirect prompt injection.
Ingestion points: The skill ingests untrusted external content using tools such as FIGMA_GET_FILE_JSON and FIGMA_GET_COMMENTS_IN_A_FILE (SKILL.md).
Boundary markers: The skill includes a 'Comment and export boundary' section which provides procedural guidelines, such as requiring explicit user authorization for posting comments, to mitigate accidental obedience to instructions found in design files.
Capability inventory: The skill utilizes tools with write capabilities, including FIGMA_ADD_A_COMMENT_TO_A_FILE, and file operations like FIGMA_DOWNLOAD_FIGMA_IMAGES (SKILL.md), which could be targeted by adversarial content within Figma nodes.
Sanitization: While the skill mentions validating temporary URLs and protecting credentials, it does not describe specific sanitization or filtering logic for the textual content retrieved from Figma assets.

figma-automation