amazon-sp-listing-optimizer
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [COMMAND_EXECUTION]: The skill requires the agent to execute local Python scripts (
validate_image.pyandstage_artifact.py) to perform its primary functions of image validation and staging. These scripts are invoked via shell commands to process media assets. - [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes product data retrieved from the Amazon Selling Partner API. Although it includes explicit instructions to detect and handle hijacked or tampered content, the ingestion of this external data remains a potential attack surface for indirect prompt injection.
- [DATA_EXFILTRATION]: In its Google Drive staging workflow, the skill modifies file permissions to make listing images public ('anyone with the link'). While documented as a requirement for Amazon's media ingestion fetcher, this results in the public exposure of the staged assets.
Audit Metadata