blog-schema
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from blog posts, FAQs, and metadata to generate structured JSON-LD, creating a surface for potential instruction hijacking.
- Ingestion points: Reads content from user-specified files (e.g.,
<file-path>) as indicated in theargument-hintand Step 1 of the workflow. - Boundary markers: The instructions do not define specific delimiters or warnings for the agent to ignore embedded instructions within the input blog content during the extraction phase.
- Capability inventory: The skill is focused on text generation and does not demonstrate dangerous capabilities like arbitrary command execution, sensitive file system access, or direct network exfiltration.
- Sanitization: The skill includes robust output sanitization instructions in Step 9, mandating the use of a real JSON encoder and the escaping of script tags (e.g., replacing
</with<\/) to prevent XSS vulnerabilities in the generated markup.
Audit Metadata