creating-agents

Design Claude Code subagents with scope, tool access, and lifecycle hooks.

Updated Jan 20, 2026
One-click install
npx skills add https://github.com/adbutterfield/coding-agent-plugins --skill creating-agents-adbutterfield
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: creating-agents
Source: https://github.com/adbutterfield/coding-agent-plugins/tree/main/plugins/meta-skills/skills/creating-agents
Command: npx skills add https://github.com/adbutterfield/coding-agent-plugins --skill creating-agents-adbutterfield

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Facilitates the design and creation of focused Claude Code subagents by guiding scope selection, tool access controls, and lifecycle hooks to ensure safe, reusable automation.

Core Features & Use Cases

  • Guides users through requirements gathering, location decisions (project vs user), and writing clear, constraint-aware descriptions.
  • Configures tool access with allowlists and permission modes to enforce least privilege and safety.
  • Adds hooks for validation, setup, and cleanup to ensure robust agent lifecycles.

Quick Start

Create a new subagent by adding a YAML frontmatter file under .claude/agents with a descriptive name and initial description.

Frequently Asked Questions about creating-agents

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I create Claude Code subagents for task-specific workflow automation?

You create Claude Code subagents by adding a YAML frontmatter file under the .claude/agents directory, defining the scope, tool access permissions, and lifecycle hooks for task-specific workflow automation.

What are lifecycle hooks in Claude Code agent design?

Lifecycle hooks in Claude Code agent design are optional configurations for validation, setup, and cleanup that ensure robust agent execution and safe automation throughout the subagent's operational lifespan.

How do I restrict tool access for subagents to enforce least privilege?

You restrict tool access for subagents by configuring tool allowlists and permission modes in the YAML frontmatter, enforcing least privilege and safety by limiting the tools available during automated workflows.

Should I deploy subagents at the project level or user level?

Choosing between project-level or user-level subagent deployment depends on your automation scope; project-level targets specific repository workflows while user-level provides broader access across multiple tasks and environments.

When should Claude delegate tasks to subagents instead of handling them directly?

Claude should delegate tasks to subagents when the workflow requires specific tool restrictions, focused scope, or structured lifecycle hooks that ensure safe, reusable automation beyond direct handling.

What is the best way to maintain focused Claude Code subagents?

The best way to maintain focused Claude Code subagents is to apply constraint-aware descriptions, enforce scope selection, and utilize frontmatter-based configuration with optional hooks for structured guidance.