Agent Development Framework FAQs

Question 1

What is the recommended model setting for sub-agents?

Accepted Answer

The framework recommends using 'model: inherit', which ensures the sub-agent utilizes the same model as the parent process for consistency and predictable behavior.

Question 2

What are the naming conventions for Claude Code agents?

Accepted Answer

Agent identifiers must be between 3-50 characters, use only lowercase letters, numbers, and hyphens, and must start and end with an alphanumeric character.

Question 3

What is a Claude Code agent compared to a standard command?

Accepted Answer

Agents are autonomous subprocesses designed for complex, multi-step tasks that require independent reasoning, whereas commands are typically user-initiated actions for specific, immediate outputs.

Question 4

How does Claude know when to trigger a specific sub-agent?

Accepted Answer

Claude uses the 'description' field in the agent's frontmatter, which includes specific triggering conditions and multi-shot examples that demonstrate the context and intent required for activation.

Question 5

Can I limit the tools available to an autonomous agent?

Accepted Answer

Yes, you can define a 'tools' array in the agent's YAML frontmatter to restrict access to specific capabilities like 'Read' or 'Grep', following the principle of least privilege.

Agent Development Framework

Key Features

Use Cases

Agent Development Framework

Key Features

Use Cases