Agent Frameworks
The backbone of an agent: these libraries run the reason–act loop, call tools, manage state, and coordinate one or many agents. Pick based on how much structure you want — from a single typed call to a full multi-agent graph.
Pricing
License
Language
Last updated: Show all
Sort: A–Z
No tools match.
Multi-Agent Orchestration
Coordinate multiple agents, tools, and reasoning steps into one workflow.
Agno
Agno
High-performance framework and runtime (AgentOS) for multi-agent systems
AutoGen
Microsoft
Microsoft's multi-agent conversation framework (maintenance-only; new work moved to Microsoft Agent Framework)
CrewAI
CrewAI
Role-based multi-agent orchestration — assemble a crew of agents with goals and tasks.
LangChain
LangChain
The original LLM-app framework — standard model/tool interfaces, hundreds of integrations, and the create_agent prebuilt that compiles to a LangGraph agent
LangGraph
LangChain
Graph-based orchestration for stateful, multi-step agents with branching, loops, and human-in-the-loop
OpenAI Agents SDK
OpenAI
OpenAI's lightweight SDK for agents with tools, handoffs, and guardrails
smolagents
Hugging Face
Hugging Face's minimal library for agents that act by writing and running code
Structured Output & Typing
Coerce model responses into validated, typed data structures.
instructor
567 Labs
Structured, validated outputs from any LLM — define a Pydantic model and get a typed object back.
Pydantic AI
Pydantic
Type-safe agent framework from the Pydantic team — structured, validated results with any model
Prompt Optimization
Program and automatically optimize prompts instead of hand-tuning them.
DSPy
Stanford NLP
Program — don't prompt — language models; declare modules and let DSPy compile and optimize the prompts.
RAG & Data Frameworks
Frameworks for retrieval-augmented generation and data-centric agents
LlamaIndex
LlamaIndex
Data framework for building RAG pipelines and document agents over your own data