skills/microsoft/eval-guide/eval-faq/Gen Agent Trust Hub

eval-faq

Pass

Audited by Gen Agent Trust Hub on May 20, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADS
Full Analysis
  • [External Content Retrieval]: The skill is configured to fetch documentation from a predefined routing table of external URLs. These sources include official Microsoft GitHub repositories, Microsoft Learn pages, and well-known industry engineering blogs. This is a standard and informative pattern for providing up-to-date technical guidance using reputable sources.
  • [Information Ingestion Surface]: The skill incorporates text from the fetched URLs into its reasoning process to answer user questions. This ingestion point is limited to the specific URLs provided in the instructions. While the skill does not use explicit boundary markers or sanitization for this external content, the potential for indirect instruction influence is mitigated by the high reputation and professional nature of the source materials.
  • [Instructional Framework]: The skill follows a clear, multi-step process (Fetch then Answer) and enforces strict answer style rules (e.g., 3-5 sentences, mandatory citations). This structure helps maintain predictable agent behavior and ensures that the output remains focused on the intended methodology guidance.
  • [Source Attribution]: The instructions mandate specific citation formats for both Microsoft and external industry sources. This practice allows for transparency and verification of the information being presented to the user.
Audit Metadata
Risk Level
SAFE
Analyzed
May 20, 2026, 10:52 PM
Security Audit — agent-trust-hub — eval-faq