amazon-store-manager

Pass

Audited by Gen Agent Trust Hub on Mar 13, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill files consist purely of Markdown documentation and prompts. There are no scripts, binaries, or automated tasks that could execute code or modify the host system.
  • [PROMPT_INJECTION]: The skill processes user-supplied product briefs and competitor URLs to generate listing copy. While it lacks explicit boundary markers, it has no destructive capabilities (such as tool-calling or shell access) that would make indirect prompt injection a risk. Ingestion points: User-provided product descriptions and external URLs. Boundary markers: Absent. Capability inventory: Limited to text generation; no tool access or execution capabilities. Sanitization: None detected.
  • [DATA_EXFILTRATION]: No hardcoded secrets, credentials, or sensitive file paths were found. The skill operates on product information provided directly by the user or found in public listings.
  • [REMOTE_CODE_EXECUTION]: No remote code execution patterns or unverifiable dependencies were detected. The skill does not use package managers or download scripts at runtime.
  • [EXTERNAL_DOWNLOADS]: The skill does not perform automated external downloads. References to external tools and URLs in the documentation are strictly informational and intended for manual use by a human operator.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 13, 2026, 03:15 AM