subagent-authoring

Create a subagent by defining YAML frontmatter with required fields (name, description, tools, model), followed by a clear system prompt that specifies the agent's role, process, and output guidelines. This structure enables Claude Code to recognize and invoke your custom agent.

A system prompt should define the agent's role, explicit invocation triggers, processing logic, tool restrictions, and output format. This clarity ensures consistent behavior and helps Claude Code properly route requests to your subagent.

Yes. Subagents support multi-file organization with references, scripts, and assets linked from the main agent file. This modular approach keeps complex agents maintainable and reusable.

Required YAML frontmatter fields include name, description, tools, and model specification. Additional best-practice fields define invocation triggers and tool restrictions to ensure your agent functions correctly within Claude Code.

Follow the structured pattern: write valid YAML frontmatter with all required fields, create a detailed system prompt with explicit triggers and tool boundaries, organize supporting files as references, and test invocation flow before deployment.

Yes. Subagent YAML frontmatter lets you specify which tools are available and define constraints in the system prompt. This gives you fine-grained control over your agent's capabilities and behavior.