linkfox-keepa-product-search
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted product listings from external sources which could act as an indirect prompt injection vector.
- Ingestion points: External data enters the agent context via responses from the
/keepa/productSearchAPI endpoint, processed inscripts/keepa_product_search.pyand saved locally. - Boundary markers: Missing; there are no explicit delimiters or strict isolation guidelines wrapping the retrieved product details to prevent the LLM from executing instructions hidden inside fields like titles or descriptions.
- Capability inventory: The skill performs local file operations (writing search results to the current directory) and external network requests to the vendor gateway via
urllibandrequestsinscripts/keepa_product_search.pyandscripts/onboarding.py. - Sanitization: Missing; retrieved product titles and descriptions are parsed as standard JSON and passed to stdout or stored without filtering for embedded instructions.
Audit Metadata