ecommerce-ozon-brand-products
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes a local Python script (scripts/ozon_brand_products.py) to facilitate communication with the NexScope API. This script is invoked with JSON parameters to perform data retrieval and processing for marketplace research.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests external marketplace data from the Ozon API, creating a surface for potential indirect prompt injection if the source data were to contain malicious instructions.
- Ingestion points: External market data is fetched via HTTP POST requests and stored in JSON format within the project's subdirectories (SKILL.md, scripts/ozon_brand_products.py).
- Boundary markers: The skill does not implement specific delimiters or 'ignore' instructions for the fetched data, relying instead on the agent's ability to parse structured JSON.
- Capability inventory: The skill has permissions to write files to the project directory for caching and data storage, and network access to the vendor's API proxy.
- Sanitization: Employs standard JSON serialization and suggests using targeted extraction tools like jq to manage the amount of external content loaded into the LLM context.
Audit Metadata