run-agent

To execute agent-specific prompts without repeating boilerplate, use a one-command invocation that automatically runs the content of @.[agent_name]/prompt/.prompt.md. This streamlines workflow automation by eliminating the need to manually reference the prompt file for each task session.

The best way to run per-agent prompt instructions is calling a single command that resolves the targeted agent via the .[agent_name] placeholder. This ensures the executed instructions are specifically tailored to the currently active coding agent.

To invoke shell tooling while disabling model invocation, execute the agent prompt file using a standardized command entry point. This satisfies the specific requirement to run shell commands without triggering model invocation during the task session.

No, you do not need to manually reference the prompt.md file each time. A one-command invocation automatically targets @.[agent_name]/prompt/.prompt.md, providing a consistent workflow entry point for composing and executing common agent tasks.

Yes, the shortcut works with different agent names by using the .[agent_name] placeholder. It dynamically resolves the executed instructions for whichever coding agent is currently targeted, ensuring agent-specific behavior across multiple task sessions.

Using a single command to execute standardized agent prompts removes the need to repeatedly restate the same instruction file. It provides an operational shortcut that creates a consistent workflow entry point for composing and executing common coding agent tasks.