audit-technical-seo
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill fetches external, untrusted content from robots.txt, XML sitemaps, and HTTP headers to generate structured audit reports and engineering tickets. This constitutes an indirect prompt injection surface where external data could attempt to manipulate agent output.
- Ingestion points: Processes for bot access audit and sitemap validation in SKILL.md involve fetching external URLs.
- Boundary markers: The skill mentions OWASP ASI01 and ASI07 guardrails to prevent goal hijacking and ensure machine-validated communication, though it lacks explicit instructions for the agent to use delimiters when processing raw external text.
- Capability inventory: The agent is authorized to use file system tools (write_file, edit_file, create_file) to produce the audit findings and tickets.
- Sanitization: While the skill provides high-level security instructions, it does not detail specific sanitization or escaping steps for the raw fetched content before it is processed by the LLM.
Audit Metadata