figma-swiftui
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted design data from Figma (via tool outputs) and user-provided SwiftUI source code. \n
- Ingestion points:
get_design_contexttool output (containing Figma JSON, React+Tailwind code, and Code Connect snippets) and local.swiftfiles provided by the user. \n - Boundary markers: The skill contains explicit instructions to treat tool output as a structural reference or design metadata instead of authoritative instructions. The
SKILL.mdfile also includes an anti-pattern warning against trusting tool responses as instructions. \n - Capability inventory: The agent can write to the local filesystem (SwiftUI code generation) and use the
use_figmatool to modify Figma files. \n - Sanitization: The instructions enforce semantic mapping to system patterns and tokens rather than allowing arbitrary execution or literal translation of external content. \n- [COMMAND_EXECUTION]: The skill instructs the agent to download image assets using
curl -Lfrom temporary asset URLs provided by the Figma MCP server to populate the local Xcode asset catalog. \n- [EXTERNAL_DOWNLOADS]: Fetches design assets from temporary endpoints hosted on the Figma MCP bridge at runtime to include them in the generated SwiftUI project.
Audit Metadata