The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external content (HTML/DOM) to extract and audit metadata tags. This constitutes an indirect prompt injection surface if malicious instructions were embedded in the processed page.
Ingestion points: The rendered_dom or raw_html of target pages provided via the environment.
Boundary markers: The skill does not define specific delimiters or instructions to ignore embedded content within the processed HTML.
Capability inventory: Uses Read, Grep, Glob, WebFetch, and Bash to perform checks and assertions against the audited content.
Sanitization: No explicit sanitization or filtering of the ingested HTML is described before auditing.
[COMMAND_EXECUTION]: The skill utilizes Bash to execute a header_check utility. This tool is specifically intended to verify the reachability, HTTP status code (200), and image dimensions of social media assets referenced in metadata.

seo-social-cards