fix-design-system-finding
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill instructions emphasize a highly constrained workflow for Figma modifications, prioritizing minimal read/write scopes and human verification for ambiguous tasks.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data from Figma nodes and audit findings, representing a standard functional surface for its intended purpose.
- Ingestion points: Ingests JSON finding data and Figma node properties retrieved through tool calls (SKILL.md).
- Boundary markers: Requires the agent to normalize and restate the finding context (fileKey, nodeId) before performing any operations, acting as a manual validation step.
- Capability inventory: Uses Figma MCP tools to perform specific structural updates (swapComponent, use_figma), though these are limited by instruction to the specific node identified in the finding.
- Sanitization: Implements safety through mandatory node backups and visual validation steps to ensure modifications match the intended fix.
Audit Metadata