craft-explore
Pass
Audited by Gen Agent Trust Hub on Nov 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a shell command using headless Chrome to render screenshots of generated design boards. Evidence found in
references/adapters/html.mdsuggesting the use ofchrome --headless=new. - [DYNAMIC_EXECUTION]: The skill generates temporary design artifacts in HTML format (
current.html,index.html, etc.) at runtime which are then rendered by external tools. This behavior is documented in the HTML adapter workflow. - [INDIRECT_PROMPT_INJECTION]: The skill defines an attack surface where instructions or data from a project's
DESIGN.mdfile or user requests are ingested and processed without explicit sanitization. - Ingestion points: Project root file
DESIGN.mdand user-provided design descriptions in chat. - Boundary markers: Absent; the skill does not specify delimiters or instructions to ignore potential malicious content within the design brief.
- Capability inventory: Write access to the local file system (HTML/CSS files), Figma/Paper MCP tool invocations, and shell execution of rendering tools.
- Sanitization: Absent; there is no validation or escaping of the design brief content before it influences the agent's output.
Audit Metadata