create-agent

Create custom AI agents with YAML frontmatter in the busytown framework.

43|4|Updated Feb 7, 2026
One-click install
npx skills add https://github.com/gordonbrander/busytown --skill create-agent-gordonbrander
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: create-agent
Source: https://github.com/gordonbrander/busytown/tree/main/.claude/skills/create-agent
Command: npx skills add https://github.com/gordonbrander/busytown --skill create-agent-gordonbrander

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill provides comprehensive guidance on creating custom agents for the busytown framework, enabling users to extend its functionality with their own specialized AI workers.

Core Features & Use Cases

  • Agent Definition: Learn how to structure agent markdown files with YAML frontmatter for persona and event subscriptions.
  • Event Handling: Understand how agents listen for specific event patterns and react to them.
  • Tool Permissions: Configure which Claude Code tools agents can access, ensuring secure and controlled operations.
  • Event Emission: Master the process of emitting events to facilitate inter-agent communication and system observability.
  • Claiming Events: Implement robust event claiming mechanisms for coordinated agent workflows.
  • Use Case: Develop a new agent that monitors code.commit events, claims them, performs a security scan using allowed tools, and emits code.vulnerability.found events if issues are detected.

Quick Start

Create a new agent by defining its name, description, and listen patterns in a markdown file within the agents/ directory.

Frequently Asked Questions about create-agent

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

FAQPage Schema
How do I build a custom AI agent in the busytown framework?

To build a custom AI agent in the busytown framework, create a markdown file in the `agents/` directory and define its `name`, `description`, persona, and event subscriptions using YAML frontmatter.

How do agents communicate and coordinate workflows in an event-driven system?

Agents communicate in an event-driven system by listening for specific event patterns, claiming events to prevent conflicts, and emitting new events to trigger downstream agents and maintain system observability.

How do I configure tool permissions for a custom agent?

You configure tool permissions for a custom agent within its YAML frontmatter, specifying which Claude Code tools the agent can access to ensure secure and controlled operations during execution.

What is event claiming and how does it work for multi-agent coordination?

Event claiming is a mechanism for multi-agent coordination that prevents multiple agents from processing the same event, ensuring robust and conflict-free execution of coordinated agent workflows.

Can I create an agent that automatically performs security scans on code commits?

Yes, you can create an agent that monitors `code.commit` events, claims them, performs a security scan using allowed tools, and emits `code.vulnerability.found` events if issues are detected.

What do I need to set up before defining a new agent?

Before defining a new agent, you need the busytown multi-agent coordination framework installed and a designated `agents/` directory to place your markdown configuration files.