ai-seo
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's primary function involves analyzing external web content, which introduces a potential surface for indirect prompt injection if the processed content contains malicious instructions.
- Ingestion points: The agent is instructed to perform visibility audits and extractability checks on user-provided domains and URLs (SKILL.md, references/agent-readiness.md).
- Boundary markers: The instructions do not specify any delimiters or safety prompts to isolate content fetched from external sources from the agent's core logic.
- Capability inventory: The skill utilizes common agent tools for analytics (SKILL.md) but does not contain any file-write, subprocess, or local execution capabilities.
- Sanitization: There are no explicit instructions for filtering or sanitizing external content.
- [EXTERNAL_DOWNLOADS]: The skill recommends the use of external packages and diagnostic scripts for specific audit tasks.
- Evidence: The reference documentation suggests the user run 'npx is-agentic' (references/agent-readiness.md) and provides a 'curl' loop snippet for diagnostic checks (references/linkedin-ai-citations.md).
- Context: These are standard diagnostic tools and instructions for the user to execute manually rather than automated processes by the agent. The mentioned services, such as Vercel and Frase, are well-known and considered safe.
Audit Metadata