The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use the agt CLI tool to apply configurations and manage worker resources. This execution pattern allows the agent to perform administrative tasks on the underlying infrastructure.
[DATA_EXFILTRATION]: The skill is designed to expose internal worker HTTP services to the public internet. As highlighted in the 'Important Notes', no authentication is configured on these exposed routes by default, creating a potential risk for unauthorized access to internal data if sensitive services are exposed.
[PROMPT_INJECTION]: The skill defines a command structure that interpolates user-provided worker names and port numbers directly into shell commands, creating an indirect prompt injection surface.
Ingestion points: User-supplied values for the --name and --expose flags in the agt CLI command examples.
Boundary markers: None. The skill does not provide instructions to delimit or treat these inputs as untrusted data.
Capability inventory: The skill has the capability to execute shell commands via the agt tool.
Sanitization: None. The documentation does not specify any validation, escaping, or filtering for the user-supplied strings before they are incorporated into a command execution context.

service-publishing