create-agent

Bootstrap modular AI agents using the OpenRouter SDK with extensible hooks and items-based streaming.

Updated Jul 6, 2026
One-click install
npx skills add https://github.com/shirulot/codex-skill --skill create-agent-shirulot
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: create-agent
Source: https://github.com/shirulot/codex-skill/tree/main/create-agent
Command: npx skills add https://github.com/shirulot/codex-skill --skill create-agent-shirulot

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires @openrouter/sdk, zod, eventemitter3, ink, react, tsx.

What problem does it solve?

This skill addresses the complexity of building scalable, modular AI agents by providing a standardized architecture that separates agent logic from UI concerns.

Core Features & Use Cases

  • Modular Architecture: Decouples the agent core from interfaces like CLI, TUI, or HTTP APIs.
  • Items-Based Streaming: Implements a robust streaming pattern that replaces items by ID, ensuring efficient UI state management.
  • Extensible Hooks: Provides a comprehensive event-driven system for logging, analytics, and custom business logic integration.

Quick Start

Initialize a new agent project by running the npm commands to install the OpenRouter SDK and setting up your TypeScript configuration.

Frequently Asked Questions about create-agent

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

FAQPage Schema
How do I build modular AI agents with OpenRouter SDK?

To build modular AI agents with OpenRouter SDK, use this skill to bootstrap a standalone agent core that decouples business logic from interfaces, utilizing items-based streaming and extensible hooks for event management.

What is items-based streaming for AI agent UI state?

Items-based streaming is a pattern that replaces streaming items by ID, ensuring efficient UI state management. It allows interfaces like Ink TUI or HTTP servers to dynamically update content without duplicating or losing track of previous responses.

Can I use TypeScript and Zod for AI agent tool definition?

Yes, you can use TypeScript and Zod for AI agent tool definition. The skill provides a standardized architecture that supports robust event management and dynamic model discovery while validating tool schemas via Zod.

Does this architecture support integrating AI agents with Discord bots?

Yes, the architecture supports integrating AI agents with Discord bots. The decoupled agent core is designed to connect with various interfaces including Ink TUI, HTTP servers, and Discord bot frameworks seamlessly.

What's the best way to separate agent logic from CLI and TUI interfaces?

The best way to separate agent logic from CLI and TUI interfaces is to bootstrap a standalone core using this skill, which utilizes an extensible event-driven hook system to handle logging and custom logic away from UI concerns.