The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to read and process local marketing context files and data from third-party SEO tools, which constitutes an ingestion surface for untrusted data.
Ingestion points: SKILL.md (Product Marketing Context section) and references/agent-readiness.md. The agent is specifically told to read files like .agents/product-marketing.md and related context files.
Boundary markers: Absent; there are no instructions to use specific delimiters or to disregard potential instructions embedded within the marketing context data.
Capability inventory: The agent is equipped to use tools for competitive analysis and data retrieval, such as semrush, ahrefs, gsc, and ga4.
Sanitization: Absent; the skill does not provide mechanisms for validating or escaping the contents of the ingested marketing files or tool outputs.
[EXTERNAL_DOWNLOADS]: The skill mentions the use of is-agentic, a package for auditing site readiness. This tool is developed by Vercel, a well-known service, and the reference is presented neutrally for diagnostic use.
[COMMAND_EXECUTION]: Documentation in references/agent-readiness.md includes the shell command npx is-agentic for performing agent-readiness audits on a domain.

ai-seo