m365-cross-app-collaboration

Pass

Audited by Gen Agent Trust Hub on Jun 27, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides high-level workflow instructions and examples for coordinating tasks across Microsoft 365 applications (Excel, PowerPoint, Word, and Outlook). It does not include any executable code, shell commands, or dangerous automation scripts.
  • [EXTERNAL_DOWNLOADS]: Includes a reference link to an official blog post from a well-known AI service (Anthropic). This link is provided for documentation and informational purposes only and does not trigger any automated downloads or background code execution.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external data, such as email attachments and document content.
  • Ingestion points: Reads content from Outlook emails, Word documents, and Excel spreadsheets.
  • Boundary markers: None explicitly defined in the instructions.
  • Capability inventory: The skill facilitates updating documents, creating PowerPoint slides, and drafting emails.
  • Sanitization: None specified. However, the instructions explicitly require 'human review before sending' for any Outlook actions, mitigating the risk of automated exploitation of the agent via malicious document content.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 27, 2026, 04:54 PM
Security Audit — agent-trust-hub — m365-cross-app-collaboration