Custom Agent Frameworks
=======================

Agent providers are the foundation for advanced customization in the Agentstr SDK, allowing you to choose and integrate your preferred agentic framework for agent logic. Unlike the high-level `AgentstrAgent` class, which offers a pre-built solution for quick setup, agent providers give you the flexibility to tailor your agent's capabilities to specific needs. This guide explores the various agent providers available, helping you select the right framework for your project.

- **LangGraph**: Refer to :doc:`../../agentstr/agents/providers/langgraph` for details on the LangGraph provider.
- **DSPy**: Refer to :doc:`../../agentstr/agents/providers/dspy` for details on the DSPy provider.
- **OpenAI**: Refer to :doc:`../../agentstr/agents/providers/openai` for details on the OpenAI provider.
- **Google**: Refer to :doc:`../../agentstr/agents/providers/google` for details on the Google provider.
- **Agno**: Refer to :doc:`../../agentstr/agents/providers/agno` for details on the Agno provider.
- **Pydantic**: Refer to :doc:`../../agentstr/agents/providers/pydantic` for details on the Pydantic provider.

These providers offer diverse approaches to agent development, from structured data handling to cutting-edge AI capabilities. Choose the one that best fits your use case, or combine multiple providers to create a hybrid agent with unique strengths. For a high-level, out-of-the-box solution, consider using `AgentstrAgent` as detailed in :doc:`agentstr_agents`.

For a complete list of agent providers and their configurations, refer to :doc:`../../agentstr/agents`.

Relevant Modules
----------------

*   :doc:`../../agentstr/agents`