The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands such as whois, dig, and curl to verify name availability. It includes robust security guidelines for the agent, requiring the removal of shell metacharacters from user-provided candidates and quoting variables to prevent command injection.
[EXTERNAL_DOWNLOADS]: The agent uses curl to fetch HTTP headers from established social media platforms like Instagram, X (Twitter), and GitHub. These operations are limited to checking status codes for username availability and target well-known, trusted services.
[DATA_EXPOSURE]: A 'Content Trust Boundary' is explicitly defined, instructing the agent to discard any directives or instructions found in external data sources like WHOIS records, DNS responses, or web search results, treating them only as data for availability signals.

product-naming