Welcome to DSPy Code

Develop DSPy Applications • Optimize with GEPA

The Complete Platform for Building and Optimizing DSPy Applications

🏗️ Develop

🧬 Optimize

🚀 Deploy

📚 Learn

Learn Build Optimize Connect

🚀 Get Started ⚡ Quick Start

🎯 What is DSPy Code?

💡 Who is This For?

🔄 Complete Development + Optimization Workflow

🚀 Use Cases: When to Use DSPy Code

1. 🏗️ Starting a New DSPy Project

Perfect for:

• Building a new AI application from scratch
• Prototyping ideas quickly
• Learning DSPy fundamentals

What you get:

dspy-code
/init --fresh

✅ Complete project structure
✅ Configuration files
✅ Example programs
✅ Best practices setup
✅ Ready to code in 2 minutes

Example: "I want to build a customer support chatbot with sentiment analysis and automated responses."

Start Your First Project →

2. 🧬 Optimizing DSPy Programs with GEPA

Perfect for: - Improving accuracy of existing programs - Automatic prompt engineering - Production optimization

What you get:

# Generate training data
→ Generate 100 examples for sentiment analysis

# Optimize automatically
/optimize my_program.py training_data.jsonl

✅ Real GEPA execution (not just code generation)
✅ Automated metric functions
✅ Progress tracking & resumption
✅ Production-ready optimized code
✅ Performance improvements documented

Real results: 75% → 92% accuracy automatically!

Example: "My sentiment analyzer is 75% accurate. Optimize it with GEPA to reach 90%+"

Learn GEPA Optimization →

3. Adding DSPy to Existing Projects

Perfect for:

• Enhancing existing Python applications
• Adding AI capabilities to current systems
• Modernizing legacy code

What you get:

cd my-existing-project
dspy-code
/init

✅ Minimal setup (no disruption)
✅ Scans your existing code
✅ Understands your project structure
✅ Generates code that fits your style
✅ Works alongside your current code

Example: "I have a Django app. I want to add AI-powered document summarization."

Add to Existing Project →

4. Learning DSPy (No Docs Required!)

Perfect for:

• First time using DSPy
• Understanding DSPy concepts
• Exploring different patterns

How it works:

Just ask questions in natural language:

→ What is a DSPy Signature?
→ How does ChainOfThought work?
→ Show me an example of ReAct
→ When should I use GEPA optimization?

The CLI answers using YOUR installed DSPy version and provides working code examples!

No reading required. Learn by building.

Start Learning →

5. Connecting to MCP Servers for Powerful DSPy Programs

DSPy Code is an MCP Client!

Connect to any MCP (Model Context Protocol) server to supercharge your DSPy programs with external tools, APIs, and data sources.

What you can do:

# Add MCP server
/mcp-add web-tools --transport stdio --command "python server.py"

# Connect to server
/mcp-connect web-tools

# Use tools in your DSPy programs
→ Create a DSPy module that searches the web and summarizes results

✅ Access external tools - Web search, databases, APIs
✅ Read from data sources - Files, documents, databases
✅ Execute commands - System operations, scripts
✅ Integrate services - Third-party APIs and tools
✅ Build powerful workflows - Combine DSPy with external capabilities

Example: "Build a RAG system that uses MCP to query my company's database and generate answers."

Learn MCP Integration →

6. Building Production AI Applications

Perfect for:

• Enterprise applications
• Production deployments
• Mission-critical systems

What you get:

✅ Validated, production-ready code
✅ Best practices built-in
✅ Error handling and logging
✅ Type hints and documentation
✅ Optimized performance
✅ Export as packages

Quality score: 90+ out of 100 automatically!

Production Best Practices →

✨ Key Features

🎬 See It In Action

Quick Example

# Start DSPy Code
dspy-code

# Initialize project
→ /init --fresh

# Connect to model
→ /model

# Generate code in natural language
→ Create a sentiment analyzer with confidence scores

# Validate
→ /validate

# Save
→ /save sentiment_analyzer.py

# Generate training data
→ Generate 50 examples for sentiment analysis

# Optimize with GEPA
→ /optimize sentiment_analyzer.py training_data.jsonl

Result: Production-ready, optimized sentiment analyzer in 5 minutes!

🏃 Quick Start

pip install dspy-code

dspy-code

/init
/model
Create a [your app]

Complete Quick Start Guide →

💬 Real Workflows

Workflow 1: Complete Beginner

Day 1:
→ Install DSPy Code
→ /init --fresh
→ "What is DSPy?"
→ "Create a simple text classifier"
→ /save my_first_program.py
→ /run

Result: Working DSPy program, understanding of basics

Workflow 2: Building Production App

Week 1:
→ dspy-code /init in existing project
→ Generate signatures, modules, programs
→ Generate 200 training examples
→ Optimize with GEPA
→ Validate (95/100 quality score)
→ Export as package
→ Deploy

Result: Production-ready AI application

Workflow 3: Learning Advanced Patterns

→ "Show me how ReAct works"
→ "Create a multi-agent system"
→ "Explain GEPA optimization"
→ "Build a RAG system with custom retrieval"

Result: Deep understanding through hands-on building

🎯 Common Questions

🚀 Ready to Start?

📚 Documentation Structure

🤝 Technical Support

Need help? Check these resources:

• FAQ - Common questions answered
• Troubleshooting - Fix common issues
• GitHub Issues - Report bugs or request features