critique
Pass
Audited by Gen Agent Trust Hub on Nov 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to read and process potentially untrusted external project files, such as
DESIGN.mdand the project's source code (HTML/CSS), which creates a vulnerability surface for malicious instructions embedded in those files to influence the agent's behavior. - Ingestion points: The agent is directed to read
DESIGN.mdat the project root and the project's design system files (SKILL.md, Step 0 and 1). - Boundary markers: There are no explicit instructions or delimiters defined to prevent the agent from following malicious commands that might be hidden within the design brief or source code.
- Capability inventory: The skill has the ability to read local files, execute headless browser commands for screenshots, and interact with Figma or Paper MCP tools to read and modify design nodes.
- Sanitization: No sanitization or content validation is performed on the ingested files or design data before they are processed by the agent.
Audit Metadata