storefront-engine
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [NO_CODE]: The skill does not distribute executable script files such as .py, .js, or .sh; instead, it provides instructions and data schemas that utilize functionality provided by the Lexsis platform and standard MCP tools.
- [EXTERNAL_DOWNLOADS]: The skill interfaces with the Lexsis MCP server at mcp.trylexsis.com and references assets from Google APIs. These are trusted service endpoints for the platform's operation.
- [COMMAND_EXECUTION]: The skill defines workflows using MCP tools for storefront management and Playwright for browser automation. These are used for legitimate tasks like visual QA and layout analysis.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external URLs provided by users (Ingestion points: references/competitor-remix.md, references/extract-template.md). It employs boundary markers such as instructions to ignore content and focus only on 'structural DNA'. The capability inventory includes the Lexsis Storefront MCP and Playwright MCP. Sanitization is performed through structural-only analysis and a sandboxed environment for execution.
Audit Metadata