java-templating
Pass
Audited by Gen Agent Trust Hub on Nov 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructions and associated reference code define patterns for ingesting and rendering third-party authored content and database-sourced data, identifying an indirect prompt injection surface.
- Ingestion points: Presentation models (e.g.,
AccountPageViewModelinreferences/presentation-examples.md) are populated from external sources such as repositories and CMS content models. - Boundary markers: The skill requires the use of "contextual escaping for dynamic values at output" and explicitly treats raw markup as a "trust boundary" requiring specific sanitization (
SKILL.md). - Capability inventory: The skill's scope is confined to Java template engines (Thymeleaf, FreeMarker, etc.) for string and file-based rendering. It does not introduce additional network, shell execution, or high-privilege capabilities.
- Sanitization: Sanitization is mandated for any unescaped markup, and contextual escaping is the default for dynamic interpolation.
Audit Metadata