Skip to content

README.md

Latest commit

 

History

History
97 lines (73 loc) · 3.83 KB

README.md

File metadata and controls

97 lines (73 loc) · 3.83 KB

Framework Comparison Examples

This directory contains equivalent implementations of multi-agent patterns across different frameworks. The goal is to demonstrate that the core patterns taught in this book are universal and can be implemented in any framework.

Structure

Each framework subdirectory mirrors the main examples/ structure:

frameworks/
├── agent-framework/     # Microsoft Agent Framework
│   ├── agents/          # Basic agents, memory, structured output
│   ├── workflows/       # Sequential workflows
│   └── orchestration/   # Round-robin, handoff patterns
├── google-adk/          # Google Agent Development Kit
│   ├── agents/          # Basic agents, memory, structured output
│   ├── workflows/       # Sequential workflows
│   └── orchestration/   # Loop agent, parallel patterns
└── langgraph/           # LangGraph (LangChain's graph-based agents)
    ├── agents/          # Basic agents, memory, structured output
    ├── workflows/       # Sequential workflows
    └── orchestration/   # Round-robin, supervisor patterns

Pattern Mapping

Pattern PicoAgents Agent Framework Google ADK LangGraph
Basic agent with tools Agent + function ChatAgent + @ai_function Agent + function create_react_agent + @tool
Memory/context ListMemory ContextProvider ToolContext.state MemorySaver checkpointer
Sequential workflow Workflow.chain() SequentialBuilder SequentialAgent StateGraph + edges
Round-robin orchestration RoundRobinOrchestrator WorkflowBuilder (cyclic) LoopAgent StateGraph + conditional edges
Parallel orchestration (manual asyncio) ConcurrentBuilder ParallelAgent (manual asyncio)
Supervisor orchestration SupervisorOrchestrator (manual) (sub_agents) StateGraph + routing
Handoff orchestration (manual) HandoffBuilder (sub_agents routing) Command/Send
Structured output output_format=Model response_format=Model response_schema=Model with_structured_output

Running Examples

Microsoft Agent Framework

# Install agent-framework
pip install agent-framework[azure]

# Set environment variables
export AZURE_OPENAI_ENDPOINT="https://your-endpoint.openai.azure.com/"
export AZURE_OPENAI_API_KEY="your-key"  # Or use Azure CLI auth

# Run examples
python examples/frameworks/agent-framework/agents/basic_agent.py
python examples/frameworks/agent-framework/orchestration/round_robin.py

Google ADK

# Install Google ADK
pip install google-adk

# Set environment variable
export GOOGLE_API_KEY="your-api-key"

# Run examples
python examples/frameworks/google-adk/agents/basic_agent.py
python examples/frameworks/google-adk/orchestration/loop_agent.py

LangGraph

# Install LangGraph and LangChain OpenAI
pip install langgraph langchain-openai python-dotenv

# Option 1: Azure OpenAI (used in examples)
export AZURE_OPENAI_ENDPOINT="https://your-endpoint.openai.azure.com/"
export AZURE_OPENAI_API_KEY="your-key"

# Option 2: OpenAI directly (see README for code changes)
export OPENAI_API_KEY="your-api-key"

# Run examples
python examples/frameworks/langgraph/agents/basic_agent.py
python examples/frameworks/langgraph/workflows/sequential.py
python examples/frameworks/langgraph/orchestration/supervisor.py

Comparison Philosophy

These examples aim to:

  1. Show equivalence: Same task, same behavior, different syntax
  2. Highlight patterns: Core patterns are framework-agnostic
  3. Be practical: Runnable examples, not pseudo-code
  4. Stay focused: Only replicate what maps cleanly between frameworks

We intentionally skip framework-specific features that don't have clear equivalents.