The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill uses npx unclaimed to perform bulk domain checks. This command downloads and executes the latest version of the unclaimed package from the NPM registry at runtime, which poses a supply chain risk as the package is not pinned to a specific version and originates from a non-trusted source.
[INDIRECT_PROMPT_INJECTION]: The skill fetches and processes content from untrusted external sources, creating an indirect prompt injection surface.
Ingestion points: Step 7a uses curl to fetch the first 16KB of content from arbitrary domain landing pages. Step 10a uses agent-browser to interact with the USPTO website.
Boundary markers: The instructions lack explicit boundary markers or warnings to the agent to ignore instructions embedded in the fetched content.
Capability inventory: The agent has access to shell commands, browser control, and the ability to perform financial transactions via vercel domains buy.
Sanitization: No sanitization or validation is performed on the ingested content beyond basic text filtering with grep.
[COMMAND_EXECUTION]: The skill relies on multiple shell utilities including vercel, whois, curl, and jq. It specifically provides the vercel domains buy command, which allows the agent to trigger domain purchases and associated financial charges on the user's account.

domain