🤖 AI Agent Frameworks

A hands-on comparison of modern AI agent and multi-agent frameworks. Get started with practical examples and explore the unique features of each framework.
Report Bug · Request Feature

This repository provides a comprehensive, hands-on comparison of modern AI agent and multi-agent frameworks. Each framework is explored through practical examples, showcasing both core functionality and its unique features, capabilities, and use cases.

Framework	Version	Docs	Repository
	0.11.5
	2.5.17
Autogen	0.7.5
Claude Agent SDK	0.1.52
	1.14.2
Google ADK	1.26.0
	1.2.13
	0.2.68
	0.14.14
Agent Framework	1.0.1
OpenAI Agents SDK	0.12.5
	1.38.0
smolagents	1.24.0
	1.32.0

The repository is organized by framework, with each top-level folder containing examples, configuration, and a README.md for that framework. The examples range from simple agent tasks to more advanced scenarios, including multi-agent workflows, RAG (Retrieval-Augmented Generation), API integrations, support for state-of-the-art protocols such as A2A and MCP, and much more.

Main modules:

• ag2/
• agno/
• autogen/
• claude-agents-sdk/
• crewai/
• google-adk/
• langchain/
• langgraph/
• llama-index/
• microsoft-agent-framework/
• openai-agents-sdk/
• pydantic-ai/
• smolagents/
• strands-agents-sdk/
• study-agents-differences/

All modules use uv for dependency management (pyproject.toml + uv.lock). Always check the README.md in each module for specific setup and usage instructions.

1. Choose a framework: Navigate to the relevant folder for the agent framework you want to explore.
2. Install dependencies: Run uv sync inside the framework folder to install all dependencies.
3. Configure API keys: Copy .env.example to .env and fill in your API keys.
4. Run examples: Execute files with uv run <filename>.py.

The study-agents-differences/ folder contains comprehensive scripts and utilities for comparing frameworks on common tasks, including RAG, API integration, and multi-agent workflows. It provides:

• Unified agent interfaces for Agno, LangGraph, LlamaIndex, OpenAI, and Pydantic-AI
• Performance benchmarks measuring response time, token usage, and tool utilization
• Detailed results and analysis comparing different agent designs and tool integrations
• Interactive Streamlit UI for real-time comparison (streamlit run agent-ui.py)

All contributions are welcome! If you have suggestions for new examples, frameworks to add, or improvements to existing content, please open an issue or submit a pull request.

• All modules use uv for dependency management (pyproject.toml + uv.lock). Check each module's README.md for installation and usage.
• Install dependencies before running examples.
• Example .env.example files are provided where needed for API keys and settings.