design-to-code
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill installs the 'coderio' Node.js package from an external registry using the 'pnpm install' command within the 'scripts/' directory. This library is a vendor resource associated with the skill author.
- [COMMAND_EXECUTION]: The skill executes multiple shell commands for environment configuration, dependency installation, and project scaffolding. It also runs a local helper script ('scripts/coderio-skill.mjs') to interact with the Figma API and manage the design-to-code workflow.
- [INDIRECT_PROMPT_INJECTION]: The skill fetches design specifications and images from the Figma API and incorporates this data into prompts for generating React components.
- Ingestion points: Figma design documents and thumbnail images are ingested into the agent context via the helper script.
- Boundary markers: The skill uses prompt templates to structure AI input but does not include explicit instructions for the AI to ignore instructions potentially embedded within the Figma design nodes.
- Capability inventory: The skill possesses the capability to write JSON data and React source code (TypeScript/TailwindCSS) to the local filesystem.
- Sanitization: The helper script utilizes functions from the 'coderio' library to simplify design nodes and extract JSON from AI responses, though it does not document formal sanitization of external design content against adversarial injection.
Audit Metadata