Building Effective Agents with Ruby

A collection of Anthropic-inspired examples showcasing how to build AI Agents with Ruby.

This repository contains implementations of common agent workflows discussed in Anthropic's blog post, Building Effective Agents.

The code here is primarily a rewrite of the sample code that was originally written in Python by Erik Schluntz and Barry Zhang here.

Installation

Clone the repo and run bundle install to install dependencies.

bundle install

Usage

These examples use OpenRouter API for LLM calls. You will need to set up an OpenRouter API key by creating a .env file using the .env.example as a template.

If you would like to use a different LLM provider, you can modify the llm_utils.rb file to use a different LLM provider along with the model of your choice as well. In these examples, the meta-llama/llama-4-scout model is used.

The LLM calls are made using the RubyLLM gem. To see the complete list of models, you can go here.

In the main.rb file, you will see examples of different agent workflows. Simply uncomment the example you want to run as such:

# Only uncomment the example you want to run
# 
# Example 1: Prompt Chaining workflow for data processing
require_relative 'examples/prompt_chaining'
# 
# Example 2: Parallelization workflow for stakeholder impact analysis
# require_relative 'examples/parallelization'
# 
# Example 3: Routing workflow for support ticket routing
# require_relative 'examples/routing'
# 
# Example 4: Orchestrator pattern with worker specialization
# require_relative 'examples/orchestrator_workers'
# 
# Example 5: Evaluator-Optimizer pattern for code generation and optimization
# require_relative 'examples/evaluator_optimizer'

Then run your selected example with ruby main.rb.

Contributing

Pull requests are always welcome.

License

MIT