amazon-category-research
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection attack surface by ingesting untrusted data from Amazon product pages that could contain malicious instructions designed to influence the agent's analysis.
- Ingestion points: Data is extracted from the
innerTextof Amazon Best Sellers pages usingmcp__playwright__browser_evaluateinSKILL.md(Steps 1.2 and 1.3). - Boundary markers: Absent. The skill does not wrap the extracted product titles, prices, or reviews in delimiters or provide instructions for the analysis agents to ignore embedded commands within the data.
- Capability inventory: The skill possesses browser automation capabilities (Playwright) to navigate to arbitrary URLs and execute JavaScript, and it can spawn background sub-agents for further processing.
- Sanitization: Absent. There is no evidence of validation, filtering, or escaping of the content scraped from the external Amazon pages before it is passed to the LLM-based analysis agents.
Audit Metadata