linkfox-walmart-search

Pass

Audited by Gen Agent Trust Hub on Oct 17, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONPERSISTENCEDATA_EXFILTRATION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill retrieves and displays product listings from Walmart, which are external and untrusted data sources.
  • Ingestion points: Product titles, descriptions, and seller names are fetched via the /walmart/search API in scripts/walmart_search.py.
  • Boundary markers: The skill does not employ explicit boundary markers or instructions to ignore embedded commands when presenting product data to the agent context.
  • Capability inventory: The skill can perform network requests to the vendor's API and write response data to local JSON files for caching and persistence.
  • Sanitization: No automated sanitization of product data was identified in the processing logic before the information is displayed.
  • [COMMAND_EXECUTION]: The skill provides instructions for the user to manually execute shell commands to configure the environment.
  • Evidence: references/onboarding.md suggests using the setx command on Windows or export on Unix-based systems to set the LINKFOX_AGENT_API_KEY.
  • [PERSISTENCE]: The onboarding documentation guides the user to modify their shell configuration files to ensure environment variables persist across sessions.
  • Evidence: Instructions in references/onboarding.md include appending export statements to ~/.zshrc or ~/.bashrc.
  • [DATA_EXFILTRATION]: The skill transmits authentication tokens and search queries to the vendor's API endpoints.
  • Evidence: Network requests are made to tool-gateway.linkfox.com and api.linkfox.com as part of the tool's standard operation and onboarding flow. These endpoints are owned by the skill vendor.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 17, 2026, 03:10 AM