The Agent Skills Directory

Command Execution Framework: The skill is designed to guide an AI agent in using the Vercel CLI (vercel or vc). This involves executing shell commands to manage deployments, environment variables, and project configurations.
[Evidence]: The skill uses bashPatterns and reference files (e.g., references/deployment.md) that contain numerous CLI command examples.
Indirect Prompt Injection Surface: The skill processes user requests via the $ARGUMENTS placeholder in the command template. While this is a standard interface for agent skills, it represents an entry point for untrusted data that could attempt to influence CLI command arguments.
[Ingestion Point]: User requests are captured within <user-request> tags in command/vercel.md.
[Boundary Markers]: The skill uses XML-style tags (<user-request>) to delimit user input from instructions.
[Capability Inventory]: The skill has the capability to execute Vercel CLI commands, which can perform network operations (deployments) and read/write local project configuration files (.vercel/).
[Sanitization]: The instructions emphasize the use of official CLI flags and recommend against risky patterns like hardcoding tokens, which provides a layer of operational safety.
Trusted External Resources: The skill references official documentation and package registries associated with Vercel, ensuring that dependencies and information come from verified sources.
[Evidence]: Links to vercel.com and instructions to install the vercel package from NPM are consistent with the vendor's official infrastructure.

vercel-cli