ecommerce-amazon-market-research

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests data from a remote API (SellerSprite via NexScope proxy), which is a standard entry point for indirect prompt injection if the remote source were compromised.
  • Ingestion points: External API responses are fetched and processed in scripts/amazon_market_research.py.
  • Boundary markers: No explicit delimiters or instructions are used to separate the external data from the agent's instructions in the script logic.
  • Capability inventory: The script performs network POST requests and writes JSON data files to the local nexscope/ directory within the workspace.
  • Sanitization: The tool acts as a data proxy and does not sanitize the JSON content before saving it to disk or displaying summaries to the user.
  • [DATA_EXPOSURE]: The skill manages authentication using the NEXSCOPE_API_KEY environment variable. It correctly directs requests to the vendor's proxy endpoint (NEXSCOPE_PROXY_BASE), following standard security practices for API-based tools.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 04:12 AM
Security Audit — agent-trust-hub — ecommerce-amazon-market-research