The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The installation instructions require the execution of an external package figma-console-mcp via npx. This package originates from a third-party developer ('southleft') that is not on the pre-approved trusted vendors list.
Evidence: npx -y figma-console-mcp@latest in the installation section.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from Figma files, which can contain untrusted content such as text nodes, comments, and component names created by external users.
Ingestion points: Tools such as figma_get_file and figma_export_tokens retrieve the full design tree and variables from external Figma files.
Boundary markers: The instructions do not provide delimiters or specific system instructions to the agent to ignore or isolate potentially malicious commands embedded within the design data.
Capability inventory: The skill provides over 100 tools, including those capable of modifying Figma files, creating nodes, and managing variables, which could be abused if the agent follows instructions found in the design content.
Sanitization: There is no evidence of sanitization or validation of the content retrieved from the Figma API before it is presented to the AI agent's context.
[COMMAND_EXECUTION]: The setup process involves running shell commands to add the MCP server and install dependencies, which executes code on the user's local environment.
Evidence: claude mcp add figma-console ... -- npx -y figma-console-mcp@latest

figma-console-mcp-design-system-api