ecommerce-1688-search-by-image

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests product data from 1688, an external source. This creates a surface for indirect prompt injection if the metadata contains malicious instructions intended to influence the agent.\n
  • Ingestion points: Product metadata returned by scripts/alibaba1688_image_search.py.\n
  • Boundary markers: The instructions lack explicit delimiters or instructions to ignore instructions embedded in search results.\n
  • Capability inventory: The skill can write to a local nexscope directory and upload local files to vendor infrastructure via scripts/upload_image.py.\n
  • Sanitization: No sanitization of the retrieved product data is implemented in the scripts.\n- [COMMAND_EXECUTION]: The skill invokes local Python scripts to process API requests and uploads. These scripts utilize standard libraries and demonstrate safe execution patterns.\n- [EXTERNAL_DOWNLOADS]: The skill performs documented network requests to vendor-managed API endpoints for searching, image uploading, and providing feedback.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 08:59 AM
Security Audit — agent-trust-hub — ecommerce-1688-search-by-image