ecommerce-amazon-alexa-search

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill connects to the NexScope API endpoint to fetch shopping data and Alexa responses. The target domain (nexscope.ai) is a recognized vendor resource, and these operations are classified as safe.
  • [COMMAND_EXECUTION]: The skill invokes a local Python script (scripts/amazon_alexa_search.py) to facilitate API interactions. The script utilizes standard libraries for HTTP requests, hashing, and JSON parsing without invoking shell commands or untrusted dynamic code.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes natural language data retrieved from the Amazon Alexa storefront. This ingestion path is a known vulnerability surface for indirect prompt injection; however, the risk is minimal given the skill's specific utility for product recommendations and shopping guidance.
  • [DATA_EXFILTRATION]: The script caches API results in a local nexscope directory within the workspace to improve performance. This behavior is restricted to the skill's own session data and does not involve accessing sensitive credentials, private keys, or system-level configuration files.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 04:12 AM
Security Audit — agent-trust-hub — ecommerce-amazon-alexa-search