store-add-product
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes shell commands using
nodefor catalog operations andnpx/npmfor verifying code integrity throughtsc --noEmitandnpm run lintwithin the target repository. - [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied product descriptions and metadata, creating an injection surface. This is addressed through strict data validation for types (integers, slugs, hex codes) and the use of TypeScript AST analysis for structural code updates, which is more secure than raw string interpolation. Evidence: Ingestion points include user-provided metadata and photo URLs; boundary markers include mandatory dry-run plans; capabilities include file writes and command execution; sanitization is performed via AST manipulation.
- [EXTERNAL_DOWNLOADS]: The instructions direct the agent to download product photos from external URLs provided by the user before processing them locally.
- [DYNAMIC_EXECUTION]: The skill dynamically generates TypeScript code blocks to represent product data and patches these into the repository's source files using the
typescriptlibrary's compiler API at runtime.
Audit Metadata