Skip to content
AgentAtlas

Best frameworks for building agents

These are the agent frameworks and nearby building blocks listed on AgentAtlas. Most are open source libraries. Dify and Flowise are visual builders you can host. Hugging Face is a hub for models and demos, not an agent library.

Microsoft's own listings say new agent projects should use Microsoft Agent Framework. AutoGen is in maintenance mode, and Semantic Kernel remains published with that same pointer.

How to choose

Choose the language first. Python libraries include LangChain, LangGraph, LlamaIndex, CrewAI, Haystack, PydanticAI, and the OpenAI Agents SDK. TypeScript listings are Mastra, the Vercel AI SDK, and Agent-Native. Google ADK ships in more than one language. Use Dify or Flowise when you want a visual editor instead of a library. Use Ollama when the missing piece is a model running on your own computer, and Composio when the agent needs signed-in access to other apps.

Python libraries

Libraries for prompts, retrieval, tools, and multi-agent flows. CrewAI's listing notes a paid cloud on top of the open source framework.

TypeScript and multi-language kits

Mastra, the Vercel AI SDK, and Agent-Native are TypeScript. Google ADK is Google's open source kit and ships in several languages.

Microsoft's libraries

Start with Microsoft Agent Framework. AutoGen and Semantic Kernel are still listed because the projects remain published.

Visual builders, models, and tool access

Dify and Flowise are open source apps with a canvas. Ollama runs models locally. Letta is for agents that keep memory. Composio connects an agent to other apps. Hugging Face hosts models, datasets, and demos.

Categories

Other guides

21 tools in this guide on AgentAtlas. Each card links to the listing, and the listing links to the official site.