m365-meeting

Pass

Audited by Gen Agent Trust Hub on Oct 16, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill makes extensive use of the m365 CLI via the Bash tool to interact with Microsoft 365 services. These operations are performed using structured command calls (e.g., m365 request) and are documented with security caveats regarding tenant policies and permissions.
  • [EXTERNAL_DOWNLOADS]: The skill facilitates downloads of meeting transcripts and recordings from official Microsoft endpoints (graph.microsoft.com and sharepoint.com). It also mentions a fallback for re-transcription using the soniox CLI, which is explicitly presented as an optional external service requiring prior user consent before any audio data is uploaded.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data in the form of meeting transcripts (VTT files). To mitigate the risk of the agent following instructions embedded within a transcript, the skill includes explicit 'operating principles' and instructions to the agent to stop after fetching the file, specifically forbidding auto-summarization or unsolicited analysis of the content.
  • [DATA_EXPOSURE]: The skill accesses sensitive organizational data such as calendar events and transcripts. It implements best practices by requiring a 'preflight' diagnostic script to verify permissions and ensuring that authentication is handled through the standard m365 login browser flow rather than insecure methods.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 16, 2026, 04:04 AM