figma-flutter-mcp-design-to-code
Warn
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: MEDIUMCREDENTIALS_UNSAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [CREDENTIALS_UNSAFE]: The skill requires a Figma API token for operation. The installation instructions guide users to pass this sensitive credential as a command-line argument (
--figma-api-key=YOUR_FIGMA_API_KEY). This practice can expose the secret in shell history, process listings, or logs. - [EXTERNAL_DOWNLOADS]: The skill setup involves downloading external resources from GitHub (
github.com/mhmzdev/figma-flutter-mcp.git) and the NPM registry. - [REMOTE_CODE_EXECUTION]: The documentation provides instructions to execute code directly from a remote source using
npx -y figma-flutter-mcp. This pattern executes code from a third-party package without prior local inspection. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest data from external Figma files, including component names, descriptions, and structures. This represents a vulnerability surface where malicious content within a Figma design could be used to manipulate the AI agent's code generation process.
- Ingestion points: Figma API responses containing design metadata and structures (e.g., frames, components, and typography definitions).
- Boundary markers: None identified in the skill instructions to separate external design data from system instructions.
- Capability inventory: The skill facilitates significant file system modifications, including creating Flutter widget files, screen layouts, and updating project configuration files like
pubspec.yaml. - Sanitization: There is no evidence of sanitization or validation of the content retrieved from Figma before it is provided as context for code generation.
Audit Metadata