Deep-Insight-Labs/awesome-ai-agents: A directory of practical and usable AI agents resources from applications and platforms to frameworks and utilities and other parts of the growing ecosystem

A curated directory of practical and usable AI agents, covering resources from applications and platforms to frameworks, utilities, and other components in the growing AI agents ecosystem.

1. Frameworks
2. Observability and Tracing
3. Platforms (Coming Soon)
4. Emerging Ideas

Frameworks provide the foundational tools for developing, orchestrating, and managing AI agents. These tools handle state management, abstractions for business logic, and other critical functionalities.

• Agentflow
  A low-code framework for creating and executing workflow defined in JSON and natural language.
  

• Agent Genesis
  An open-source framework that allows developers to build custom Retrieval-Augmented Generation (RAG) workflows and AI agents using modular, copy-paste components.
  

• AgentVerse
  A versatile platform offering tools for building and managing AI agents, particularly suited for collaborative environments.
  

• Agno
  A lightweight library for building Multimodal Agents with memory, knowledge and tools. Notably, Agno emphasizes performance and scalability – e.g. agent instantiation is benchmarked ~10,000× faster than some alternatives.
  

• Atomic Agent
  A versatile, open-source framework created by BrainBlend AI for developing multi-agent systems and AI applications.
  

• AutoGPT
  An experimental open-source application demonstrating the capabilities of GPT-4 by autonomously developing and managing projects, requiring minimal human intervention.
  

• Autogen
  A Microsoft framework designed for building multi-agent conversational systems, enabling the creation of cooperative AI agents that can autonomously generate and manage complex tasks.
  

• Bee Agent Framework
  An open-source toolkit designed to create scalable agent-based workflows with various AI models.
  

• CAMEL
  Emerging from a research community exploring the “scaling laws” of AI agents. Designed as a sandbox to study how collaborative AI systems perform and evolve as the number of agents grows.
  

• CrewAI
  A framework for orchestrating role-playing AI agents, allowing developers to define distinct roles and interactions among agents to simulate complex scenarios.
  

• Haystack
  An end-to-end framework for building NLP applications, facilitating the creation of systems that can perform tasks like question answering and document retrieval using Transformer models and LLMs.
  

• Internet of Agents
  An open-source framework enabling AI agents to collaborate and tackle complex tasks together, facilitating distributed problem-solving.
  

• LangChain
  A framework that simplifies the development of applications powered by large language models by providing tools for chaining together various components, such as prompts, models, and memory.
  

• LangGraph
  A framework that provides a visual interface for stateful, multi-actor applications with LLMs, used to create agent and multi-agent workflows.
  

• LlamaIndex
  A data framework that offers a central interface to connect your LLMs with external data sources, streamlining the process of data ingestion and retrieval for language models.
  

• MetaGPT
  A multi-agent meta programming framework that, given a single-line requirement, can autonomously generate a Product Requirement Document (PRD), design documents, tasks, repository setup, and continuous integration configurations.
  

• Modelscope-Agent
  A customizable and scalable agent framework offering capabilities such as role-playing, LLM integration, and tool usage.
  

• OpenAI Swarm
  An experimental framework by OpenAI’s dev team for lightweight multi-agent orchestration. Introduces concepts of “agents”, “handoffs” and “routines”. Agents encapsulate a task or expertise, and handoffs allow an agent to pass control to another agent. Routines are reusable plans that an agent can replay to fulfill certain objectives.
  

• Pydantic AI
  A new experimental framework that integrates Pydantic’s structured validation with AI agent workflows. Pydantic AI focuses on enforcing strong type safety, schema validation, and structured output generation for AI-driven applications.
  

• Semantic Kernel
  A Microsoft SDK that integrates large language models with conventional programming languages, enabling developers to incorporate AI capabilities into their applications seamlessly.
  

• smolagents
  A minimalist library by Hugging Face that enables developers to build powerful AI agents with a few lines of code, focusing on code agents that write and execute Python code snippets for tasks.
  

• SuperAGI
  An open-source framework enabling developers to build, manage, and run autonomous AI agents efficiently, supporting concurrent agent operations and tool integration.
  

This section will include tools and resources to monitor, debug, and analyze AI agent behaviors and interactions.

• AgentOps
  A platform designed to monitor multi-agent systems, providing insights into performance bottlenecks and collaboration quality, thereby enhancing efficiency and reliability.
  

• Explorer by Invariant Labs
  An advanced observability tool designed to simplify the analysis of agent traces, allowing users to quickly browse and understand agent behaviors and interactions.
  

• Helicone AI
  An open-source observability platform designed to empower developers in building, monitoring, and optimizing AI applications, providing real-time insights and analytics.
  

• Langfuse
  A tool that enables monitoring, tracing, and debugging of AI agents, assisting developers in optimizing application performance and understanding agent behaviors.
  

• LangSmith
  An all-in-one developer platform for every step of the LLM-powered application lifecycle, offering capabilities to debug, collaborate, test, and monitor LLM applications.
  

Stay tuned for updates! This section will cover platforms that enable you to deploy and manage AI agents in production environments.

• Model Context Protocol
  Open protocol proposed by Anthropic to standardize integration between LLM applications and external data sources and tools.

Have a framework, platform, or tool that you think belongs here? Feel free to open a pull request or share your recommendations.

🌟 Help us grow this directory by starring the repositories and sharing this page! Together, we can accelerate innovation in the AI agent space. 🌟