The Agent Skills Directory

[PERSISTENCE]: The configuration reference in references/figma-mcp-config.md instructs users to add environment variable exports to shell profile files like ~/.zshrc or ~/.bashrc to persist the FIGMA_OAUTH_TOKEN across sessions.
[INDIRECT_PROMPT_INJECTION]: The skill processes structured design data retrieved from external Figma URLs, which could contain malicious instructions embedded by third parties in design layers.
Ingestion points: Untrusted data enters the agent context via the get_design_context, get_metadata, and get_figjam tools which fetch content from user-provided Figma links.
Boundary markers: The skill does not define specific delimiters or "ignore embedded instructions" warnings for the Figma payload.
Capability inventory: The skill uses the fetched data to generate React and Tailwind code, and can download assets based on the payload.
Sanitization: There is no documented validation or sanitization of the Figma node content before it is processed for code generation.
[SAFE]: The skill connects to the official Figma MCP endpoint (https://mcp.figma.com/mcp), which is a well-known service infrastructure.

figma