The Agent Skills Directory

[COMMAND_EXECUTION]: The skill generates and executes JavaScript snippets within the Figma plugin environment using the figma-use tool (documented in references/sub-agent-prompts.md). These scripts programmatically traverse Figma's node tree to extract structured design data.
[DATA_EXFILTRATION]: When operating in extension mode, the skill performs extensive read operations on the local codebase, including model definitions (app/models/, prisma/schema.prisma), routing configurations (config/routes.rb), and environment templates (.env.example). This behavior is intended for codebase analysis but represents a significant data exposure surface.
[PROMPT_INJECTION]: The skill processes untrusted content from external Figma designs to generate technical specifications and architectural reviews, creating a vulnerability to indirect prompt injection.
Ingestion points: Figma design metadata and JSON data retrieved via figma-dev and figma-use MCP tools.
Boundary markers: The instructions lack explicit boundary markers or delimiters to isolate content retrieved from Figma from the agent's internal instructions.
Capability inventory: The agent possesses the ability to read local application source code and write multiple Markdown files to the local docs/ directory.
Sanitization: No specific sanitization, filtering, or validation of the text extracted from Figma nodes is performed before it is interpolated into the generated documentation.

figma-spec-extractor