umbraco-review-content-model
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill operates in a read-only capacity, gathering information about Umbraco content models via Model Context Protocol (MCP) tools or local markdown files. It does not perform any destructive actions or modify the target environment.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests external data in the form of site schema (names, aliases, descriptions) and requirements documents. Although this data is untrusted, the skill's functionality is limited to generating a markdown report, which mitigates the risk of instructions embedded in the schema being executed. 1. Ingestion points: Umbraco Developer MCP tool outputs, requirements markdown files in the
docs/umbraco-schema/directory, and user-provided JSON schema exports. 2. Boundary markers: Analysis results are interpolated into a predefined markdown report template (assets/review-report.md). 3. Capability inventory: Uses MCP tools for data retrieval and writes a report file todocs/; instructs the agent to run a local script (lint-requirements.mjs) for validation. 4. Sanitization: No explicit sanitization is defined for schema metadata before report interpolation, relying on the agent's context handling.
Audit Metadata