semi-design-guide

Pass

Audited by Gen Agent Trust Hub on Oct 29, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill references the installation of the official MCP tool @douyinfe/semi-mcp via npx, which is a resource provided by the skill's author.
  • [EXTERNAL_DOWNLOADS]: The documentation in WORKFLOWS.md suggests using standard web testing tools such as @midscene/web and playwright for component verification.
  • [COMMAND_EXECUTION]: A script template is provided to execute UI tests using Playwright, which is a common practice for verifying generated frontend code.
  • [INDIRECT_PROMPT_INJECTION]: The skill introduces a workflow for UI verification where an AI agent interacts with a web page based on natural language instructions.
  • Ingestion points: The verify-ui.mjs script processes the visual content of a local project preview defined by the PREVIEW_URL variable.
  • Boundary markers: The testing template does not specify clear delimiters between user instructions and page content.
  • Capability inventory: The testing agent has the capability to perform actions (aiAct) and make assertions (aiAssert) on the rendered page.
  • Sanitization: No specific sanitization of the page content is described before it is interpreted by the AI agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 29, 2026, 07:47 AM