msgraph-integration-patterns

Pass

Audited by Gen Agent Trust Hub on Sep 5, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [Indirect Prompt Injection Surface]: The skill details methods for retrieving data from external sources, specifically Microsoft Teams chats, channel messages, and SharePoint transcripts. As this data originates from third parties, it represents an ingestion point for untrusted content. If this data is subsequently interpreted by an AI agent without adequate delimiters or sanitization, it could serve as a vector for indirect prompt injection attacks.
  • Ingestion points: Microsoft Graph endpoints (/me/chats, /teams/{teamId}/channels/{channelId}/messages) and SharePoint REST API.
  • Boundary markers: The patterns focus on API integration and do not specify prompt-level delimiters for the retrieved content.
  • Capability inventory: The environment allows the use of bash, web_fetch, write_file, and edit_file tools.
  • Sanitization: The patterns emphasize technical ID validation rather than content-based sanitization for LLM safety.
  • [External Library Reference]: The skill references @azure/msal-browser via a well-known module delivery service. This is a standard method for implementing authentication using official resources from a trusted vendor.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 5, 2026, 04:28 AM
Security Audit — agent-trust-hub — msgraph-integration-patterns