The Agent Skills Directory

[PROMPT_INJECTION]: The skill is designed to fetch and process content from external, untrusted websites for analysis. This creates a surface for indirect prompt injection, where an attacker could place malicious instructions within a website's HTML, metadata, or robots.txt to influence the agent's behavior. \n
Ingestion points: The skill uses tools like web_fetch and curl to retrieve external site data as part of the technical SEO and crawlability audits described in SKILL.md.\n
Boundary markers: There are no explicit instructions or delimiters used to tell the agent to ignore instructions embedded within the fetched data.\n
Capability inventory: The agent possesses network fetching capabilities and access to local context (e.g., .agents/product-marketing-context.md), which could be targeted by an injection attack.\n
Sanitization: The skill does not define any sanitization or validation logic for the content it retrieves from remote sources. \n- [EXTERNAL_DOWNLOADS]: The skill fetches data from remote URLs provided by the user or discovered during the audit (e.g., robots.txt, XML sitemaps, and page content). While essential for the skill's function, this involves processing data from external domains.

seo-audit