schema
Pass
Audited by Gen Agent Trust Hub on Jun 14, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by instructing the agent to ingest external context files, such as .agents/product-marketing.md, without implementing boundary markers or instructions to disregard embedded commands. 1. Ingestion points: SKILL.md specifies reading .agents/product-marketing.md and .claude/product-marketing.md. 2. Boundary markers: No delimiters or ignore-instructions are provided for external content. 3. Capability inventory: The skill is limited to generating JSON-LD structured data blocks. 4. Sanitization: There is no validation or sanitization of the ingested content.
- [EXTERNAL_DOWNLOADS]: The skill references established and trusted validation tools from Google (search.google.com) and Schema.org (validator.schema.org), which is standard and safe practice for structured data implementation.
Audit Metadata