ecommerce-amazon-ads-manager
Pass
Audited by Gen Agent Trust Hub on Aug 31, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes a local Python script (scripts/amazon_api.py) to process API requests and interact with the Nexscope gateway. The script relies exclusively on Python standard libraries and implements defensive measures such as response size limits (32 MiB) to prevent resource exhaustion.\n- [INDIRECT_PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection by ingesting advertising data from external API responses via the Nexscope gateway.\n
- Ingestion points: Data is fetched and processed in scripts/amazon_api.py from the external advertising API gateway.\n
- Boundary markers: Instructions in SKILL.md mandate a human-in-the-loop verification process, requiring the agent to present a preview of any mutation and wait for explicit user approval before execution.\n
- Capability inventory: The skill has the capability to perform authenticated mutation operations (POST/PUT) on Amazon Advertising entities, including campaigns and targeting rules.\n
- Sanitization: The script performs JSON validation and unwrapping, though it does not filter the natural language content of API responses for potential adversarial instructions.
Audit Metadata