wb-connector

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill retrieves user-authored content (product descriptions, reviews, and questions) from the Wildberries marketplace, which could contain malicious directives.
  • Ingestion points: Data enters the agent's context through tools like wb_search, wb_reviews, and wb_questions.
  • Boundary markers: The skill explicitly defines a 'Trust boundary' section that instructs the agent to treat fetched strings as untrusted data and specifically ignore commands such as 'ignore previous instructions'.
  • Capability inventory: The skill's toolset is limited to querying public marketplace data. It does not provide the agent with capabilities for file system modification, persistent storage, or arbitrary network access.
  • Sanitization: The skill documentation notes automated handling of character encoding (auto-decoding latin-1 to utf-8) to ensure data integrity.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 09:00 PM
Security Audit — agent-trust-hub — wb-connector