create-agent

Guide creation of Claude Code subagents with YAML frontmatter and system prompts.

Updated Feb 17, 2026
One-click install
npx skills add https://github.com/filipowm/claude-plugins --skill create-agent-filipowm
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: create-agent
Source: https://github.com/filipowm/claude-plugins/tree/main/plugins/claude-helper/skills/create-agent
Command: npx skills add https://github.com/filipowm/claude-plugins --skill create-agent-filipowm

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This Skill guides users through the entire process of designing and creating a new custom Claude Code subagent, from initial concept to a fully defined agent file.

Core Features & Use Cases

  • Interactive Design: Guides users through 8 distinct phases to define agent identity, behavior, workflow, scope, capabilities, and runtime configuration.
  • Automated File Generation: Creates the .md agent file with appropriate YAML frontmatter and system prompt based on user input.
  • Scope Management: Supports project-level, user-level, and plugin-level agent creation, with logic to detect plugin roots.
  • Use Case: A user wants to create a specialized agent to review code for security vulnerabilities. This Skill will walk them through defining its name, expertise, workflow, tool access, model choice, and finally generate the agent file in the correct location.

Quick Start

Use the create-agent skill to design a new agent named 'code-security-auditor' with the purpose of finding security flaws in code.

Frequently Asked Questions about create-agent

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

FAQPage Schema
How do I create a custom Claude Code subagent for a specific task?

Creating a custom Claude Code subagent involves interactively defining its identity, role, behavior, and capabilities. The system then automatically generates a complete agent `.md` file with YAML frontmatter and a structured system prompt.

What is the process for designing a Claude agent's workflow and scope?

Designing a Claude agent's workflow and scope requires going through eight distinct interactive phases. These phases capture the agent's expertise, tool access, runtime configuration, and model choice to output a fully defined agent file.

Can I configure subagents at the project, user, or plugin level?

Yes, you can configure subagents at the project, user, or plugin level. The creation process supports all three scopes and includes logic to automatically detect plugin roots for accurate file placement.

How do I generate a Claude agent file with YAML frontmatter and a system prompt?

You generate a Claude agent file by providing the agent's design specifications interactively. The system uses your input to automatically produce the `.md` file with the appropriate YAML frontmatter and a structured system prompt.

What's the best way to define tool access and model choice for a custom agent?

The best way to define tool access and model choice is through the interactive design phases for custom agents. These phases guide you in specifying runtime configurations and capabilities directly before file generation.