This repository provides side-by-side examples of AI agent implementations across various frameworks, enabling developers to compare approaches and use them as project templates. It targets developers building AI agents who need to evaluate different frameworks and best practices for common LLM workflows.

How It Works

The project showcases identical agent functionalities, such as a customer support agent, implemented using multiple AI agent frameworks. Each example uses the same Google Gemini 2.5 Flash model and undergoes scenario tests with the same model for verification. This approach allows for direct comparison of framework design philosophies and implementation patterns for identical tasks.

Quick Start & Requirements

• Install dependencies: uv sync --all-groups
• Run tests: uv run pytest -s tests/test_customer_support_agent.py
• Debug mode: uv run pytest -s tests/test_customer_support_agent.py --debug
• Requires Python.
• Official documentation and demo links are not explicitly provided in the README.

Highlighted Details

• Compares Agno, DSPy, Google ADK, InspectAI, LangGraph (Functional & High-level APIs), Pydantic AI, and smolagents.
• Aims to cover LLM workflows and agent examples from Anthropic's "Building Effective Agents" guide.
• Currently includes Customer Support Agent and Vibe Coding Agent examples.
• Actively seeking contributions for new frameworks and agent use cases.

Maintenance & Community

• The project is maintained by langwatch.
• A Discord server is available for discussions.
• Users are encouraged to open issues for requests and contributions.

Licensing & Compatibility

• License: MIT
• The MIT license permits commercial use and integration into closed-source projects.

Limitations & Caveats

The repository is a work in progress, with several agent examples and frameworks listed as "coming soon" or "pending." The focus is on demonstrating core agent patterns rather than providing production-ready, feature-complete implementations for each framework.