seo-meta-onpage
Pass
Audited by Gen Agent Trust Hub on Sep 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external HTML content from potentially untrusted sources for SEO auditing purposes.
- Ingestion points: Processes
rendered_dom,raw_html, or content retrieved viaWebFetchfrom external URLs. - Boundary markers: The skill relies on specific tags and a structured findings schema to guide the agent, which helps isolate the data from the prompt instructions.
- Capability inventory: Uses
Bashto execute a local parsing script andWebFetchto access remote sites. - Sanitization: The use of a dedicated parsing script (
parse-html.mjs) to return structured data (title, length, etc.) instead of passing raw HTML directly to the agent's context serves as a significant security control.
Audit Metadata