Does it require Sorbet?

Yes, dspy-ruby leverages Sorbet to provide type-safe signatures, ensuring that data passed to and from the LLM adheres to your application's schema.

dspy-ruby is a Ruby framework inspired by DSPy for building LLM applications using programmatic modules and signatures instead of manual string prompts.

How does prompt optimization work in this skill?

It uses advanced algorithms like GEPA (Genetic-Pareto Reflective Prompt Evolution) to analyze LLM outputs against metrics and automatically refine instructions for better performance.

Can I integrate dspy-ruby with Ruby on Rails?

Absolutely. It includes conventions for directory structures, lifecycle hooks, and storage systems designed specifically for seamless Rails integration.

Which LLM providers are supported?

It supports major providers including OpenAI, Anthropic, and Gemini via dedicated adapter gems or the unified RubyLLM interface.

DSPy for Ruby

Name: DSPy for Ruby
Author: gvkhosla

bygvkhosla

•

データサイエンスとML

Builds type-safe, programmatic LLM applications in Ruby using structured signatures and automated prompt optimization.

DSPy.rb brings software engineering rigor to LLM development by replacing manual prompt tweaking with modular, type-safe Ruby components. By leveraging Sorbet for schema definitions and advanced optimization algorithms like GEPA and MIPROv2, it allows developers to build predictable AI features, complex agentic workflows, and production-ready integrations that improve automatically based on data rather than guesswork. This skill is essential for Ruby developers moving beyond basic API wrappers into sophisticated, maintainable AI system architecture.

主な機能

01Deep Rails integration with lifecycle callbacks and organized directory structures

02Type-safe Signatures using Sorbet for rigorous input/output validation

03Structured event system for observability and performance monitoring

04Automatic prompt optimization via GEPA and MIPROv2 algorithms

0535 GitHub stars

06Modular components including ChainOfThought and ReAct for complex logic

ユースケース

01Building autonomous agents with type-safe tool use in Ruby applications

02Optimizing prompt performance using genetic algorithms and feedback loops

03Creating predictable LLM-powered classifiers or data extraction pipelines

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add gvkhosla/compound-engineering-pi dspy-ruby

For use in Claude.ai and ChatGPT

Download Skill