figma-flutter-mcp-design-to-code

Warn

Audited by Gen Agent Trust Hub on May 21, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSPROMPT_INJECTIONCREDENTIALS_UNSAFE
Full Analysis
  • [UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: The skill directs users to execute code from a non-trusted source using 'npx' and to clone a repository from an unverified third-party GitHub user ('mhmzdev').
  • [DATA_EXPOSURE_AND_EXFILTRATION]: To function, the skill requires a Figma Personal Access Token. This token is passed to the third-party MCP server, which is an unverified external tool.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes content from Figma files, which are external and untrusted ingestion points. Maliciously crafted Figma designs could attempt to influence or override agent instructions.
  • Ingestion points: External Figma API data via design links.
  • Boundary markers: None provided in instructions.
  • Capability inventory: File system writes for asset export and 'pubspec.yaml' updates.
  • Sanitization: No sanitization process is described for the Figma data.
  • [REMEDIATION]: Ensure Figma tokens are only used with trusted local tools. Review Figma design files from unknown sources before processing them with AI agents to prevent injection attacks.
Audit Metadata
Risk Level
MEDIUM
Analyzed
May 21, 2026, 09:08 AM
Security Audit — agent-trust-hub — figma-flutter-mcp-design-to-code