The Agent Skills Directory

[SAFE]: The skill implementation follows a clear, non-malicious logic for importing design specifications. It interacts with Figma, an established and well-known service, using specific data extraction tools. No indicators of credential theft, remote code execution, or persistence were detected.\n- [INDIRECT_PROMPT_INJECTION]: The skill has a data ingestion surface as it processes external Figma layer names, metadata, and variables provided by users. This is the skill's primary intended function. The risk is minimized because the skill's capabilities are limited to internal data aggregation and handoff to subsequent agent steps.\n
Ingestion points: Figma node metadata, layer hierarchies, and variable definitions retrieved through tools like get_metadata and get_variable_defs described in SKILL.md.\n
Boundary markers: Absent; there are no explicit delimiters or instructions provided to the agent to disregard instructions embedded within the design data.\n
Capability inventory: Data extraction is restricted to generating an internal "extraction record" for use by the figma-build skill; no tools for arbitrary file access, shell execution, or network exfiltration are present.\n
Sanitization: No sanitization or escaping mechanisms for the retrieved design text are specified in the instructions.

figma-import