umbraco-mcp-content-ops

Pass

Audited by Gen Agent Trust Hub on Mar 29, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The instructions focus on guiding an agent through legitimate content management tasks in Umbraco CMS using specified tool collections.
  • [SAFE]: Includes explicit safety rules in SKILL.md and references/guardrails.md that require user confirmation for destructive actions (delete, move to recycle bin) and bulk operations (affecting more than 25 items).
  • [SAFE]: Encourages security best practices such as checking user permissions before performing mutations and validating data using official Umbraco endpoints.
  • [SAFE]: The skill defines a clear scope that excludes high-risk operations like schema changes or infrastructure management by default.
  • [SAFE]: Regarding indirect prompt injection (Category 8): 1. Ingestion points: The skill ingests data from local files (CSV/JSON) via filesystem MCP and content from the Umbraco content tree. 2. Boundary markers: No explicit instruction-ignoring delimiters are defined. 3. Capability inventory: Actions include content CRUD, publishing, and media management within the Umbraco environment. 4. Sanitization: Relies on Umbraco's internal validation endpoints and mandatory user confirmation for high-impact changes, providing sufficient mitigation for the intended use case.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 29, 2026, 05:18 PM
Security Audit — agent-trust-hub — umbraco-mcp-content-ops