The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill requires the agent to perform extensive research and analysis of external, untrusted data sources to evaluate search visibility.
Ingestion points: The agent is instructed to analyze competitor content, third-party review sites (G2, Capterra), forum discussions (Reddit, Quora), and live AI search results across multiple platforms.
Boundary markers: The skill lacks explicit instructions or delimiters to prevent the agent from obeying malicious instructions that might be embedded in the external pages it is tasked with auditing.
Capability inventory: The skill uses SEO tools (Semrush, Ahrefs, GSC) and potentially file system operations to audit the user's project files (e.g., robots.txt, .agents/product-marketing.md).
Sanitization: No methods for sanitizing or filtering untrusted external content are provided in the instructions.
[EXTERNAL_DOWNLOADS]: The skill references and suggests the use of various third-party utilities and monitoring services to track AI visibility.
Evidence: The documentation recommends the is-agentic Node.js package and references several external monitoring services including Otterly AI, Peec AI, and ZipTie.

ai-seo