同步完整源码 - 2026-05-25
This commit is contained in:
xiaoxue
commit
0e55297929
|
|
@ -0,0 +1,69 @@
|
|||
# Contributing to Awesome Agent Orchestration
|
||||
|
||||
Thank you for considering contributing to this repository! This document provides guidelines for contributing.
|
||||
|
||||
## Ways to Contribute
|
||||
|
||||
1. **Add new resources** - Share frameworks, tools, or papers
|
||||
2. **Improve existing entries** - Update descriptions or add missing information
|
||||
3. **Fix errors** - Correct typos or outdated information
|
||||
4. **Translate** - Add translations to other languages
|
||||
|
||||
## Contribution Guidelines
|
||||
|
||||
### Adding New Resources
|
||||
|
||||
When adding a new resource, please ensure:
|
||||
|
||||
1. **Relevance**: The resource must be related to agent orchestration, multi-agent systems, or swarm intelligence
|
||||
2. **Quality**: Prefer well-maintained, actively developed projects
|
||||
3. **Format**: Follow the existing table format in each section
|
||||
4. **Stars Badge**: Use the shields.io format:
|
||||
```
|
||||
|
||||
```
|
||||
|
||||
### Table Format
|
||||
|
||||
For framework/tool entries:
|
||||
```markdown
|
||||
| [Project Name](URL) | Description |  |
|
||||
```
|
||||
|
||||
For research papers:
|
||||
```markdown
|
||||
| [Paper Title](URL) | Description | Year |
|
||||
```
|
||||
|
||||
### Sections Available
|
||||
|
||||
- Core Frameworks
|
||||
- Multi-Agent Orchestration Frameworks
|
||||
- Swarm Intelligence Frameworks
|
||||
- Graph-Based Workflow Frameworks
|
||||
- Lightweight & Specialized Agents
|
||||
- Orchestration Patterns
|
||||
- Communication & Coordination
|
||||
- Research Papers
|
||||
- Tools & Infrastructure
|
||||
- Learning Resources
|
||||
|
||||
## Pull Request Process
|
||||
|
||||
1. Fork the repository
|
||||
2. Create a feature branch: `git checkout -b add-new-framework`
|
||||
3. Make your changes
|
||||
4. Commit with clear messages
|
||||
5. Push to your fork
|
||||
6. Submit a Pull Request
|
||||
|
||||
## Review Criteria
|
||||
|
||||
- Is the resource relevant to agent orchestration?
|
||||
- Is the information accurate and up-to-date?
|
||||
- Is the formatting consistent with existing entries?
|
||||
- Are the links working?
|
||||
|
||||
---
|
||||
|
||||
Thank you for your contributions!
|
||||
|
|
@ -0,0 +1,62 @@
|
|||
CC0 1.0 Universal
|
||||
|
||||
Statement of Purpose
|
||||
|
||||
The laws of most jurisdictions throughout the world automatically confer
|
||||
exclusive Copyright and Related Rights (defined below) upon the creator and
|
||||
subsequent owner(s) (each and all, an "owner") of an original work of
|
||||
authorship and/or a database (each, a "Work").
|
||||
|
||||
Certain works are described below in Canada as having been and are hereby
|
||||
released to the public domain under the terms of CC0 1.0 Universal Public
|
||||
Domain Dedication ("CC0 1.0 Dedication").
|
||||
|
||||
Creative Commons has designed CC0 1.0 to be a "no rights reserved" legal
|
||||
tool that allows creators and owners to dedicate their works to the public
|
||||
domain. In addition, Creative Commons makes reference to the associated
|
||||
Legal Code from which the dedication is taken, and that current version of
|
||||
the legal code is posted below.
|
||||
|
||||
Statement of Purpose
|
||||
|
||||
The laws of most jurisdictions throughout the world automatically confer
|
||||
exclusive Copyright and Related Rights (defined below) upon the creator and
|
||||
subsequent owner(s) (each and all, an "owner") of an original work of
|
||||
authorship and/or a database (each, a "Work").
|
||||
|
||||
Certain works are described below in Canada as having been and are hereby
|
||||
released to the public domain under the terms of CC0 1.0 Universal Public
|
||||
Domain Dedication ("CC0 1.0 Dedication").
|
||||
|
||||
To the extent possible under law, on behalf of myself and my successors
|
||||
in interest and assigns, I have waived and will waive, to the extent
|
||||
permitted by law, all rights related to the Work, including but not
|
||||
limited to my copyright, neighboring rights, moral rights, rights of
|
||||
publicity, rights of privacy, and rights of attribution in the Work.
|
||||
|
||||
This waiver includes my use of your organization's trademarks or brand
|
||||
features in an promotional context. It does not extend to uses that would
|
||||
infringe on your organization's trademark or other intellectual property
|
||||
rights, or on personal or moral rights. It also does not extend to any
|
||||
uses of the Work that are licensed or governed by intellectual property
|
||||
rights owned or controlled by your organization (including patent and
|
||||
trademark rights). You should review your organization's policies and
|
||||
practices, and consult with your own legal counsel, to ensure that your
|
||||
intended use of our organization's trademarks or brand features is
|
||||
permitted.
|
||||
|
||||
I acknowledge that this dedication is made with the understanding that
|
||||
the Work as contributed will be attributed to the organization or
|
||||
individual that contributed to the creation of the Work as listed in the
|
||||
required attribution section of the Work, if any. If no attribution
|
||||
section is required, the attribution shall be considered to be as
|
||||
follows:
|
||||
|
||||
"The original version of this Work was contributed by [Organization
|
||||
Name], as listed in the following source: link to source repository."
|
||||
|
||||
Theholder of copyright and associated rights in theabove-identified
|
||||
Work hereby irrevocably releases the Work into the public domain under
|
||||
the terms of CC0 1.0 Universal Public Domain Dedication.
|
||||
|
||||
Full text of CC0 1.0: https://creativecommons.org/publicdomain/zero/1.0/
|
||||
|
|
@ -0,0 +1,474 @@
|
|||
# Awesome Agent Orchestration
|
||||
|
||||
<!--
|
||||
🎯 Search Keywords: ai agent, multi-agent, agent orchestration, swarm intelligence,
|
||||
llm agent, autonomous agent, agent framework, agentic ai, crewai, autogen,
|
||||
metagpt, langgraph, swarms, a2a protocol, mcp protocol, agent workflow,
|
||||
ai automation, multi-agent system, agent communication, ai agents
|
||||
|
||||
🔥 GitHub Topics: ai-agent, multi-agent, agent-orchestration, swarm-intelligence,
|
||||
llm-agents, autogen, crewai, metagpt, langgraph, swarms, a2a-protocol, mcp
|
||||
-->
|
||||
|
||||
[English](./README.md) | [中文](./README_zh.md)
|
||||
|
||||
[](https://awesome.re)
|
||||
[](LICENSE)
|
||||
[](https://github.com/vivy-yi/awesome-agent-orchestration/stargazers)
|
||||
[](CONTRIBUTING.md)
|
||||
|
||||
**The most comprehensive curated list of Agent Orchestration, Multi-Agent Systems, Swarm Intelligence, and LLM Agent frameworks.** Includes AutoGen, CrewAI, MetaGPT, LangGraph, Swarms, A2A Protocol, MCP Protocol and more.
|
||||
|
||||
⭐ **Star this repo if it helps you find AI Agent frameworks!**
|
||||
|
||||
## Table of Contents
|
||||
|
||||
- [Why Agent Orchestration?](#why-agent-orchestration)
|
||||
- [Core Frameworks](#core-frameworks)
|
||||
- [Multi-Agent Orchestration Frameworks](#multi-agent-orchestration-frameworks)
|
||||
- [Swarm Intelligence Frameworks](#swarm-intelligence-frameworks)
|
||||
- [Graph-Based Workflow Frameworks](#graph-based-workflow-frameworks)
|
||||
- [Type-Safe & Modern Frameworks](#type-safe--modern-frameworks)
|
||||
- [TypeScript & Node.js Frameworks](#typescript--nodejs-frameworks)
|
||||
- [Go Frameworks](#go-frameworks)
|
||||
- [Swift, Ruby & Elixir Frameworks](#swift-ruby--elixir-frameworks)
|
||||
- [PHP & Java Frameworks](#php--java-frameworks)
|
||||
- [Agent Communication Protocols](#agent-communication-protocols)
|
||||
- [Agent-to-Agent (A2A)](#agent-to-agent-a2a)
|
||||
- [Model Context Protocol (MCP)](#model-context-protocol-mcp)
|
||||
- [Orchestration Patterns](#orchestration-patterns)
|
||||
- [Hierarchical](#hierarchical)
|
||||
- [Sequential](#sequential)
|
||||
- [Parallel/Concurrent](#parallelconcurrent)
|
||||
- [Mixture of Agents (MoA)](#mixture-of-agents-moa)
|
||||
- [Agent Council](#agent-council)
|
||||
- [Communication & Coordination](#communication--coordination)
|
||||
- [Research Papers](#research-papers)
|
||||
- [Tools & Infrastructure](#tools--infrastructure)
|
||||
- [Learning Resources](#learning-resources)
|
||||
- [Related Awesome Lists](#related-awesome-lists)
|
||||
|
||||
---
|
||||
|
||||
## What is Agent Orchestration?
|
||||
|
||||
**Agent Orchestration** is the practice of coordinating multiple AI agents (LLM agents, autonomous agents) to work together on complex tasks. It enables:
|
||||
|
||||
- **Scalability**: Handle complex workflows by distributing tasks across specialized agents
|
||||
- **Specialization**: Each agent can focus on a specific domain or task
|
||||
- **Robustness**: Fault tolerance through agent redundancy and recovery
|
||||
- **Emergent Capabilities**: Complex behaviors arising from simple agent interactions
|
||||
- **Real-World Applications**: From software development to research automation
|
||||
|
||||
### Related Concepts
|
||||
|
||||
- **Multi-Agent System (MAS)**: A system composed of multiple interacting intelligent agents
|
||||
- **Swarm Intelligence**: Collective behavior emerging from simple agent interactions
|
||||
- **Agentic AI**: AI systems that can autonomously plan and execute complex tasks
|
||||
- **Agent Communication Protocol**: Standards like A2A (Agent-to-Agent) and MCP (Model Context Protocol) for agent interoperability
|
||||
|
||||
---
|
||||
|
||||
## Core Frameworks
|
||||
|
||||
### Multi-Agent Orchestration Frameworks
|
||||
|
||||
| Project | Description | Stars |
|
||||
|---------|-------------|-------|
|
||||
| [AutoGen](https://github.com/microsoft/autogen) | Microsoft's framework for building multi-agent conversational systems |  |
|
||||
| [CrewAI](https://github.com/crewAIInc/crewAI) | Role-based multi-agent framework for building AI teams |  |
|
||||
| [MetaGPT](https://github.com/geekan/MetaGPT) | Multi-agent framework for software development with role assignment |  |
|
||||
| [agentUniverse](https://github.com/agentuniverse-ai/agentUniverse) | Enterprise-level multi-agent framework |  |
|
||||
| [Langroid](https://github.com/langroid/langroid) | Multi-agent programming framework with message-passing |  |
|
||||
| [Phidata](https://github.com/agno-agi/phidata) | Multi-modal agent framework with built-in tools and UI |  |
|
||||
| [OpenAgents](https://github.com/OpenAgentsInc/openagents) | Open platform for deploying language agents |  |
|
||||
| [AgentVerse](https://github.com/OpenBML/AgentVerse) | Platform for running multiple LLM agents in simulated environments |  |
|
||||
| [agency-swarm](https://github.com/VRSEN/agency-swarm) | Reliable multi-agent orchestration framework |  |
|
||||
| [OpenAI Swarm](https://github.com/openai/swarm) | Educational lightweight multi-agent orchestration framework |  |
|
||||
| [Microsoft Agent Framework](https://github.com/microsoft/agent-framework) | Framework for building, orchestrating and deploying AI agents |  |
|
||||
| [OpenAI Agents Python](https://github.com/openai/openai-agents-python) | Lightweight powerful framework for multi-agent workflows |  |
|
||||
| [Haystack](https://github.com/deepset-ai/haystack) | Open-source AI orchestration for context-engineered LLM applications |  |
|
||||
| [LLMStack](https://github.com/trypromptly/LLMStack) | No-code multi-agent framework for LLM agents and workflows |  |
|
||||
| [LazyLLM](https://github.com/LazyAGI/LazyLLM) | Easiest way for building multi-agent LLM applications |  |
|
||||
| [MindSearch](https://github.com/InternLM/MindSearch) | LLM-based multi-agent framework for web search engine |  |
|
||||
| [TradingAgents](https://github.com/TauricResearch/TradingAgents) | Multi-agent LLM financial trading framework |  |
|
||||
| [Shannon](https://github.com/Kocoro-lab/Shannon) | Production-oriented multi-agent orchestration framework |  |
|
||||
| [Solace Agent Mesh](https://github.com/SolaceLabs/solace-agent-mesh) | Event-driven framework for multi-agent AI systems |  |
|
||||
| [Pipelex](https://github.com/Pipelex/pipelex) | Declarative language for composable AI workflows |  |
|
||||
| [CodeFuse-muAgent](https://github.com/codefuse-ai/CodeFuse-muAgent) | Innovative agent framework driven by KG Engine |  |
|
||||
| [L2MAC](https://github.com/samholt/L2MAC) | LLM Automatic Computer Framework |  |
|
||||
| [Patchwork](https://github.com/patched-codes/patchwork) | Agentic AI framework for enterprise workflow automation |  |
|
||||
|
||||
### Swarm Intelligence Frameworks
|
||||
|
||||
| Project | Description | Stars |
|
||||
|---------|-------------|-------|
|
||||
| [Swarms](https://github.com/kyegomez/swarms) | Production-grade multi-agent infrastructure platform |  |
|
||||
| [Swarm](https://github.com/autoreason/swarm) | Lightweight multi-agent orchestration framework |  |
|
||||
| [Swarm-Tools](https://github.com/FelipeDaza7/swarm-tools) | Tool for coordinating AI agents with learning capabilities |  |
|
||||
| [Ruflo](https://github.com/ruvnet/ruflo) | Leading agent orchestration platform for Claude |  |
|
||||
| [swarms-rs](https://github.com/The-Swarm-Corporation/swarms-rs) | Enterprise-grade multi-agent orchestration in Rust |  |
|
||||
| [Lux](https://github.com/Spectral-Finance/lux) | Open-source framework for multi-agent swarmed intelligence |  |
|
||||
| [AutoRT](https://github.com/kyegomez/AutoRT) | Embodied foundation models for robotic agents |  |
|
||||
|
||||
### Graph-Based Workflow Frameworks
|
||||
|
||||
| Project | Description | Stars |
|
||||
|---------|-------------|-------|
|
||||
| [LangGraph](https://github.com/langchain-ai/langgraph) | Build stateful multi-agent applications with LangChain |  |
|
||||
| [LangChain](https://github.com/langchain-ai/langchain) | Modular framework for LLM application chaining |  |
|
||||
| [LlamaIndex](https://github.com/jerryjliu/llama_index) | Data-centric agent framework for RAG |  |
|
||||
| [Semantic Kernel](https://github.com/microsoft/semantic-kernel) | Microsoft orchestration SDK for .NET/Python |  |
|
||||
|
||||
### Lightweight & Specialized Agents
|
||||
|
||||
| Project | Description | Stars |
|
||||
|---------|-------------|-------|
|
||||
| [SmolAgents](https://github.com/smol-ai/smolagents) | Minimalist library for code-writing agents |  |
|
||||
| [Auto-GPT](https://github.com/Significant-Gravitas/Auto-GPT) | Pioneering autonomous agent with self-prompting |  |
|
||||
| [BabyAGI](https://github.com/yoheinakajima/babyagi) | Simple autonomous task management |  |
|
||||
| [GPT-Engineer](https://github.com/AntonOsika/gpt-engineer) | Agent for generating full projects from specs |  |
|
||||
| [BMTools](https://github.com/OpenBMB/BMTools) | Tool/API registry for agents |  |
|
||||
|
||||
### Type-Safe & Modern Frameworks
|
||||
|
||||
| Project | Description | Stars |
|
||||
|---------|-------------|-------|
|
||||
| [PydanticAI](https://github.com/pydantic/pydantic-ai) | Type-safe agent framework using Pydantic models |  |
|
||||
| [DeepAgents](https://github.com/vstorm-co/pydantic-deepagents) | Production-grade autonomous agents on PydanticAI |  |
|
||||
| [Subagents](https://github.com/vstorm-co/subagents-pydantic-ai) | Subagent delegation framework for PydanticAI |  |
|
||||
| [Flux0](https://github.com/flux0-ai/flux0) | Multi-agent deployment framework with session management |  |
|
||||
| [Mamba Agents](https://github.com/sequenzia/mamba-agents) | Simple extensible AI Agent framework |  |
|
||||
|
||||
### TypeScript & Node.js Frameworks
|
||||
|
||||
| Project | Description | Stars |
|
||||
|---------|-------------|-------|
|
||||
| [openai-swarm-node](https://github.com/youseai/openai-swarm-node) | Node.js implementation of OpenAI's Swarm framework |  |
|
||||
| [Network-AI](https://github.com/jovanSAPFIONEER/Network-AI) | TypeScript/Node multi-agent orchestrator with shared state |  |
|
||||
| [megazord](https://github.com/Sh3rd3n/megazord) | Claude Code framework for multi-agent coordination |  |
|
||||
| [Voltmachines](https://github.com/ssdeanx/Voltmachines) | Multi-agent framework built on VoltAgent |  |
|
||||
| [agent-swarm-kit](https://github.com/tripolskypetr/agent-swarm-kit) | TypeScript library for framework-agnostic multi-agent systems |  |
|
||||
| [ai-orchestra](https://github.com/langtail/ai-orchestra) | Lightweight orchestration for AI Agents around Vercel's streamText |  |
|
||||
|
||||
### Go Frameworks
|
||||
|
||||
| Project | Description | Stars |
|
||||
|---------|-------------|-------|
|
||||
| [openai-agents-go](https://github.com/nlpodyssey/openai-agents-go) | Lightweight powerful framework for multi-agent workflows in Go |  |
|
||||
| [swarm-go](https://github.com/feiskyer/swarm-go) | Ergonomic lightweight multi-agent orchestration in Go |  |
|
||||
| [rs-graph-llm](https://github.com/a-agmon/rs-graph-llm) | High-performance framework for interactive multi-agent workflows in Rust |  |
|
||||
|
||||
### Swift, Ruby & Elixir Frameworks
|
||||
|
||||
| Project | Description | Stars |
|
||||
|---------|-------------|-------|
|
||||
| [SwiftSwarm](https://github.com/jamesrochabrun/SwiftSwarm) | Swift framework for lightweight multi-agent orchestration |  |
|
||||
| [Swarm](https://github.com/christopherkarani/Swarm) | Lightweight agent orchestration framework in Swift |  |
|
||||
| [ruby-openai-swarm](https://github.com/graysonchen/ruby-openai-swarm) | Ruby-based framework for multi-agent orchestration |  |
|
||||
| [swarm_ex](https://github.com/nrrso/swarm_ex) | Elixir library for lightweight AI agent orchestration |  |
|
||||
| [shifts](https://github.com/aaronrussell/shifts) | Elixir framework for composing autonomous AI agent workflows |  |
|
||||
|
||||
### PHP & Java Frameworks
|
||||
|
||||
| Project | Description | Stars |
|
||||
|---------|-------------|-------|
|
||||
| [neuron-ai](https://github.com/neuron-core/neuron-ai) | PHP agentic framework for production-ready AI applications |  |
|
||||
| [swarm-ai](https://github.com/intelliswarm-ai/swarm-ai) | Java multi-agent orchestration with Spring AI |  |
|
||||
|
||||
### Workflow & Pipeline Frameworks
|
||||
|
||||
| Project | Description | Stars |
|
||||
|---------|-------------|-------|
|
||||
| [circuit](https://github.com/smogili1/circuit) | Drag-and-drop agent workflow builder |  |
|
||||
| [routilux](https://github.com/lzjever/routilux) | Event-driven workflow orchestration for Python |  |
|
||||
| [CraftFlow](https://github.com/scholarlords/CraftFlow) | Workflow orchestration for RAG and multi-agent collaborations |  |
|
||||
| [sentinels](https://github.com/garyblankenship/sentinels) | Laravel package for agent-based task orchestration |  |
|
||||
| [titan-orchestrator](https://github.com/ramn51/titan-orchestrator) | Distributed orchestrator bridging DevOps and Agentic AI |  |
|
||||
|
||||
---
|
||||
|
||||
## Agent Communication Protocols
|
||||
|
||||
### Agent-to-Agent (A2A)
|
||||
|
||||
Google's A2A protocol enables secure, interoperable communication between autonomous agents across frameworks and vendors.
|
||||
|
||||
| Project | Description | Stars |
|
||||
|---------|-------------|-------|
|
||||
| [awesome-a2a-agents](https://github.com/isekOS/awesome-a2a-agents) | Curated list of A2A tools and frameworks |  |
|
||||
| [python-a2a](https://github.com/themanojdesai/python-a2a) | Python implementation of A2A protocol |  |
|
||||
| [a2a-net](https://github.com/neuroglia-io/a2a-net) | .NET implementation of A2A protocol |  |
|
||||
| [a2a-langgraph](https://github.com/ruska-ai/a2a-langgraph) | A2A protocol built on LangGraph |  |
|
||||
| [nestjs-a2a](https://github.com/thestupd/nestjs-a2a) | NestJS library for A2A protocol |  |
|
||||
| [a2a4j](https://github.com/PheonixHkbxoic/a2a4j) | Java implementation of A2A protocol |  |
|
||||
| [A2A-MCP-Server](https://github.com/GongRzhe/A2A-MCP-Server) | Bridge between MCP and A2A protocols |  |
|
||||
| [a2a-x402](https://github.com/google-agentic-commerce/a2a-x402) | A2A protocol with cryptocurrency payments |  |
|
||||
| [Routa](https://github.com/phodal/routa) | Multi-agent coordination platform via MCP/A2A |  |
|
||||
| [mangaba_ai](https://github.com/Mangaba-ai/mangaba_ai) | Minimalist AI agents with A2A and MCP |  |
|
||||
|
||||
### Model Context Protocol (MCP)
|
||||
|
||||
An open protocol that enables seamless connection between AI models and external tools/data sources. MCP is becoming the standard for tool-augmented LLM applications.
|
||||
|
||||
#### Official Resources
|
||||
|
||||
| Project | Description | Stars |
|
||||
|---------|-------------|-------|
|
||||
| [awesome-mcp-servers](https://github.com/wong2/awesome-mcp-servers) | Most popular curated list of MCP servers |  |
|
||||
| [modelcontextprotocol/registry](https://github.com/modelcontextprotocol/registry) | Community-driven MCP server registry |  |
|
||||
| [microsoft/mcp](https://github.com/microsoft/mcp) | Official Microsoft MCP servers catalog |  |
|
||||
| [modelcontextprotocol/swift-sdk](https://github.com/modelcontextprotocol/swift-sdk) | Official Swift SDK for MCP |  |
|
||||
|
||||
#### Database & Vector Stores
|
||||
|
||||
| Project | Description | Stars |
|
||||
|---------|-------------|-------|
|
||||
| [mcp-server-qdrant](https://github.com/qdrant/mcp-server-qdrant) | Official Qdrant vector database server |  |
|
||||
| [mcp-server-milvus](https://github.com/zilliztech/mcp-server-milvus) | Milvus vector database server |  |
|
||||
| [mcp-neo4j](https://github.com/neo4j-contrib/mcp-neo4j) | Neo4j graph database server |  |
|
||||
| [mongodb-mcp-server](https://github.com/mongodb-js/mcp-mcp-server) | MongoDB database server |  |
|
||||
| [mysql_mcp_server](https://github.com/designcomputer/mysql_mcp_server) | MySQL database server |  |
|
||||
| [MariaDB/mcp](https://github.com/MariaDB/mcp) | MariaDB database server |  |
|
||||
|
||||
#### Browser & Automation
|
||||
|
||||
| Project | Description | Stars |
|
||||
|---------|-------------|-------|
|
||||
| [mcp-playwright](https://github.com/executeautomation/mcp-playwright) | Playwright browser automation |  |
|
||||
| [mcp-chrome](https://github.com/hangwin/mcp-chrome) | Chrome extension for browser control |  |
|
||||
| [mobile-mcp](https://github.com/mobile-next/mobile-mcp) | iOS/Android automation |  |
|
||||
|
||||
#### Development & Tools
|
||||
|
||||
| Project | Description | Stars |
|
||||
|---------|-------------|-------|
|
||||
| [mcp-filesystem-server](https://github.com/mark3labs/mcp-filesystem-server) | Filesystem operations |  |
|
||||
| [jupyter-mcp-server](https://github.com/datalayer/jupyter-mcp-server) | Jupyter notebook integration |  |
|
||||
| [fetch-mcp](https://github.com/zcaceres/fetch-mcp) | HTTP fetching |  |
|
||||
| [mcp-graphql](https://github.com/blurrah/mcp-graphql) | GraphQL API integration |  |
|
||||
| [mcpadapt](https://github.com/grll/mcpadapt) | Connect 650+ MCP servers to any framework |  |
|
||||
|
||||
#### Specialized Servers
|
||||
|
||||
| Project | Description | Stars |
|
||||
|---------|-------------|-------|
|
||||
| [arxiv-mcp-server](https://github.com/blazickjp/arxiv-mcp-server) | arXiv paper search |  |
|
||||
| [mcp-youtube](https://github.com/anaisbetts/mcp-youtube) | YouTube video analysis |  |
|
||||
| [excel-mcp-server](https://github.com/haris-musa/excel-mcp-server) | Excel file manipulation |  |
|
||||
| [drawio-mcp-server](https://github.com/lgazo/drawio-mcp-server) | Diagram creation |  |
|
||||
| [freecad-mcp](https://github.com/neka-nat/freecad-mcp) | FreeCAD 3D modeling |  |
|
||||
| [ida-mcp-server](https://github.com/MxIris-Reverse-Engineering/ida-mcp-server) | IDA reverse engineering |  |
|
||||
| [kubernetes-mcp-server](https://github.com/containers/kubernetes-mcp-server) | Kubernetes management |  |
|
||||
|
||||
---
|
||||
|
||||
## Orchestration Patterns
|
||||
|
||||
### Hierarchical
|
||||
|
||||
In hierarchical orchestration, a supervisor agent delegates tasks to specialized sub-agents. This pattern is suitable for complex tasks requiring domain expertise.
|
||||
|
||||
- **Use Cases**: Software development teams, research automation, enterprise workflows
|
||||
- **Frameworks**: AutoGen, CrewAI, LangGraph
|
||||
|
||||
### Sequential
|
||||
|
||||
Sequential workflows process tasks through a pipeline where each agent completes its step before passing to the next.
|
||||
|
||||
- **Use Cases**: Document processing, data transformation pipelines
|
||||
- **Frameworks**: LangChain, LangGraph
|
||||
|
||||
### Parallel/Concurrent
|
||||
|
||||
Multiple agents work simultaneously on independent subtasks, with results aggregated afterward.
|
||||
|
||||
- **Use Cases**: Research gathering, parallel code generation
|
||||
- **Frameworks**: Swarms, AutoGen
|
||||
|
||||
### Mixture of Agents (MoA)
|
||||
|
||||
Multiple expert agents contribute their perspectives, with outputs combined through iterative refinement.
|
||||
|
||||
- **Research**: [Mixture of Agents Paper](https://arxiv.org/abs/2406.04692)
|
||||
- **Implementation**: Swarms MoA, LangGraph
|
||||
|
||||
### Agent Council
|
||||
|
||||
Multiple agents discuss and debate to reach consensus on decisions.
|
||||
|
||||
- **Use Cases**: Decision making, complex analysis, creative writing
|
||||
- **Framework**: AutoGen GroupChat
|
||||
|
||||
---
|
||||
|
||||
## Communication & Coordination
|
||||
|
||||
### Message Passing
|
||||
|
||||
- **Langroid**: Message-passing based multi-agent framework
|
||||
- **AutoGen**: Conversational message passing
|
||||
|
||||
### Shared State / BlackBoard Systems
|
||||
|
||||
- **Swarm Matcher**: Agent matching and selection
|
||||
- **Message Pool**: Shared communication system
|
||||
|
||||
### Task Distribution
|
||||
|
||||
- **Auto Swarm**: Self-organizing swarm with automatic task distribution
|
||||
- **Agent Delegation**: Task delegation and management
|
||||
|
||||
---
|
||||
|
||||
## Research Papers
|
||||
|
||||
### Multi-Agent Collaboration
|
||||
|
||||
| Paper | Description | Year |
|
||||
|-------|-------------|------|
|
||||
| [CAMEL: Communicative Agents for "Mind" Exploration](https://arxiv.org/abs/2303.17760) | Role-playing agent collaboration through dialogue | 2023 |
|
||||
| [MetaGPT: Multi-Agent Collaboration Framework](https://arxiv.org/abs/2308.07370) | Software development with agent roles (PM, Engineer, QA) | 2023 |
|
||||
| [AgentVerse: Flexible Multi-Agent Platform](https://arxiv.org/abs/2308.07429) | Collaborative problem-solving in simulated environments | 2023 |
|
||||
| [Generative Agents](https://arxiv.org/abs/2304.03442) | Interactive agents with long-term memory in virtual worlds | 2023 |
|
||||
| [Mixture of Agents](https://arxiv.org/abs/2406.04692) | Parallel processing with iterative refinement | 2024 |
|
||||
|
||||
### Orchestration & Planning
|
||||
|
||||
| Paper | Description | Year |
|
||||
|-------|-------------|------|
|
||||
| [MALT: Multi-Agent LLM Orchestration](https://arxiv.org/abs/2412.01928) | Structured conversations with Creator-Verifier-Refiner pattern | 2024 |
|
||||
| [Agent-as-a-Judge](https://arxiv.org/abs/2410.10934) | Agents evaluate other agents | 2024 |
|
||||
| [ReAct: Synergizing Reasoning and Acting](https://arxiv.org/abs/2210.03629) | Interleaving thoughts and actions | 2022 |
|
||||
| [Chain-of-Thought Prompting](https://arxiv.org/abs/2201.11903) | Step-by-step reasoning in LLM prompts | 2022 |
|
||||
| [Tree-of-Thoughts](https://arxiv.org/abs/2305.10601) | Multiple reasoning branches for complex problems | 2023 |
|
||||
| [Reflexion](https://arxiv.org/abs/2309.00668) | Self-critique loop for long-horizon tasks | 2023 |
|
||||
| [HuggingGPT](https://arxiv.org/abs/2303.17580) | LLM controller orchestrating expert models | 2023 |
|
||||
| [Toolformer](https://arxiv.org/abs/2302.04761) | Self-supervised fine-tuning for API calling | 2023 |
|
||||
|
||||
### Memory & Long-Term Autonomy
|
||||
|
||||
| Paper | Description | Year |
|
||||
|-------|-------------|------|
|
||||
| [Generative Agents](https://arxiv.org/abs/2304.03442) | Interactive agents simulating humans in virtual worlds | 2023 |
|
||||
| [Survey on Memory Mechanisms of LLM Agents](https://arxiv.org/abs/2402.13736) | Overview of memory architectures and retrieval | 2024 |
|
||||
|
||||
### Evaluation
|
||||
|
||||
| Benchmark | Description |
|
||||
|-----------|-------------|
|
||||
| [AgentBench](https://arxiv.org/abs/2308.07258) | Evaluates agents across diverse environments |
|
||||
| [Multi-Agent Evaluation](https://arxiv.org/abs/2312.04323) | Survey of metrics for agent evaluation |
|
||||
|
||||
---
|
||||
|
||||
## Tools & Infrastructure
|
||||
|
||||
### Observability & Tracing
|
||||
|
||||
| Tool | Description |
|
||||
|------|-------------|
|
||||
| [Langfuse](https://github.com/langfuse/langfuse) | Open-source LLM engineering platform |
|
||||
| [Phoenix](https://github.com/Arize-ai/phoenix) | Open-source LLM tracing |
|
||||
| [Opik](https://github.com/comet-ml/opik) | Open-source LLM evaluation |
|
||||
| [Agenta](https://github.com/agenta-ai/agenta) | LLMOps platform for prototyping |
|
||||
| [Logfire](https://github.com/pydantic/logfire) | Pydantic's open-source observability tool |
|
||||
| [Langtrace](https://github.com/Scale3-Labs/langtrace) | Open-source observability for AI agents |
|
||||
| [Okahu Monocle](https://github.com/OkahuAI/monocle) | Tracing framework by Linux Foundation |
|
||||
|
||||
### Memory & Knowledge
|
||||
|
||||
| Tool | Description |
|
||||
|------|-------------|
|
||||
| [Memori](https://github.com/MemoriLabs/Memori) | SQL Native Memory Layer for LLMs & Agents |
|
||||
| [Letta](https://github.com/letta/letta) | Memory layer for AI agents |
|
||||
|
||||
### Deployment & Infrastructure
|
||||
|
||||
| Tool | Description |
|
||||
|------|-------------|
|
||||
| [OpenAgents](https://github.com/OpenAgentsInc/openagents) | Platform for deploying language agents |
|
||||
| [FastAPI Agents](https://github.com/blairhudson/fastapi-agents) | FastAPI extension for AI agent frameworks |
|
||||
| [AgentAPIProduction](https://github.com/The-Swarm-Corporation/AgentAPIProduction) | Production-ready agent API system |
|
||||
| [AgentOS](https://github.com/The-Swarm-Corporation/AgentOS) | Operating system for AI agents |
|
||||
|
||||
### Evaluation & Benchmarking
|
||||
|
||||
| Tool | Description |
|
||||
|------|-------------|
|
||||
| [AgentBench](https://github.com/AgentBench/AgentBench) | Evaluates agents across diverse environments |
|
||||
| [swarms-evals](https://github.com/The-Swarm-Corporation/swarms-evals) | Evaluation framework for swarm systems |
|
||||
| [RAGAS](https://github.com/explodinggradients/ragas) | Evaluation framework for RAG systems |
|
||||
|
||||
---
|
||||
|
||||
## Learning Resources
|
||||
|
||||
### Tutorials & Guides
|
||||
|
||||
- [LLM Powered Autonomous Agents](https://lilianweng.github.io/posts/2023-06-15-agent/) - Lilian Weng's comprehensive guide
|
||||
- [Prompt Engineering Guide: LLM Agents](https://github.com/dair-ai/Prompt-Engineering-Guide#llm-agents) - Comprehensive agent building guide
|
||||
- [Building Autonomous Agents with LangChain](https://www.youtube.com/watch?v=example) - Video tutorial
|
||||
- [Enterprise-Grade-Agents-Course](https://github.com/The-Swarm-Corporation/Enterprise-Grade-Agents-Course) - Comprehensive course
|
||||
|
||||
### Documentation
|
||||
|
||||
- [AutoGen Documentation](https://microsoft.github.io/autogen/) - Microsoft AutoGen docs
|
||||
- [CrewAI Documentation](https://docs.crewai.com/) - CrewAI official docs
|
||||
- [LangGraph Documentation](https://langchain-ai.github.io/langgraph/) - LangGraph docs
|
||||
- [Swarms Documentation](https://docs.swarms.world/) - Swarms framework docs
|
||||
|
||||
---
|
||||
|
||||
## Related Awesome Lists
|
||||
|
||||
- [awesome-ai-agents](https://github.com/heurema/awesome-ai-agents) - Comprehensive AI agent resources
|
||||
- [awesome-LangGraph](https://github.com/von-development/awesome-LangGraph) - LangChain/LangGraph ecosystem
|
||||
- [awesome-multi-agent-papers](https://github.com/kyegomez/awesome-multi-agent-papers) - Multi-agent research papers
|
||||
- [awesome-Swarms-List](https://github.com/The-Swarm-Corporation/Awesome-Swarms-List) - Swarms framework resources
|
||||
- [awesome-a2a-agents](https://github.com/isekOS/awesome-a2a-agents) - A2A protocol ecosystem
|
||||
- [awesome-agent-protocols](https://github.com/shanjai-raj/awesome-agent-protocols) - Agent communication protocols
|
||||
- [awesome-agents](https://github.com/l-aime/awesome-agents) - Cutting-edge AI agent projects
|
||||
- [awesome-agentic-ai](https://github.com/mlnjsh/awesome-agentic-ai) - Agentic AI frameworks and papers
|
||||
|
||||
---
|
||||
|
||||
## Contributing
|
||||
|
||||
Contributions are welcome! Please read the [contributing guidelines](CONTRIBUTING.md) first.
|
||||
|
||||
1. Fork the repository
|
||||
2. Add your resource in the appropriate section
|
||||
3. Ensure consistent formatting
|
||||
4. Submit a pull request
|
||||
|
||||
---
|
||||
|
||||
## SEO Keywords
|
||||
|
||||
> Popular search terms: AI Agent Framework, Multi-Agent System, Agent Orchestration, Swarm Intelligence, LLM Agent, Claude AI, OpenAI Agents, CrewAI, AutoGen, MetaGPT, LangGraph, Agent Communication Protocol, A2A Protocol, MCP Protocol, Agentic AI, Autonomous Agents, Agent Workflow, AI Automation
|
||||
|
||||
### Popular Frameworks
|
||||
|
||||
- [AutoGen](https://github.com/microsoft/autogen) - Microsoft's multi-agent framework
|
||||
- [CrewAI](https://github.com/crewAIInc/crewAI) - Role-based multi-agent framework
|
||||
- [MetaGPT](https://github.com/geekan/MetaGPT) - Software development multi-agent framework
|
||||
- [LangGraph](https://github.com/langchain-ai/langgraph) - State-based agent workflow
|
||||
- [Swarms](https://github.com/kyegomez/swarms) - Production-grade multi-agent platform
|
||||
|
||||
### Popular Protocols
|
||||
|
||||
- [A2A Protocol](https://google.github.io/A2A/) - Agent-to-Agent communication
|
||||
- [MCP](https://modelcontextprotocol.io/) - Model Context Protocol
|
||||
|
||||
---
|
||||
|
||||
## License
|
||||
|
||||
[](LICENSE)
|
||||
|
||||
To the extent possible under law, [Your Name](https://github.com/yourusername) has waived all copyright and related rights to this work.
|
||||
|
||||
---
|
||||
|
||||
*Star this repository if you find it useful!*
|
||||
|
|
@ -0,0 +1,474 @@
|
|||
# Awesome Agent Orchestration
|
||||
|
||||
<!--
|
||||
🔍 搜索关键词: ai智能体, 多智能体, 智能体编排, 群体智能,
|
||||
llm智能体, 自主智能体, 智能体框架, 智能体ai, crewai, autogen,
|
||||
metagpt, langgraph, swarms, a2a协议, mcp协议, 智能体工作流,
|
||||
ai自动化, 多智能体系统, 智能体通信, ai agents
|
||||
|
||||
🏷️ GitHub标签: ai-agent, multi-agent, agent-orchestration, swarm-intelligence,
|
||||
llm-agents, autogen, crewai, metagpt, langgraph, swarms, a2a-protocol, mcp
|
||||
-->
|
||||
|
||||
[English](./README.md) | [中文](./README_zh.md)
|
||||
|
||||
[](https://awesome.re)
|
||||
[](LICENSE)
|
||||
[](https://github.com/vivy-yi/awesome-agent-orchestration/stargazers)
|
||||
[](CONTRIBUTING.md)
|
||||
|
||||
**最全面的智能体编排、多智能体系统、群体智能和LLM智能体框架精选列表。** 包含 AutoGen、CrewAI、MetaGPT、LangGraph、Swarms、A2A 协议、MCP 协议等。
|
||||
|
||||
⭐ **如果帮到你找到 AI 智能体框架，请给这个仓库点个 Star！**
|
||||
|
||||
## 目录
|
||||
|
||||
- [为什么需要智能体编排？](#为什么需要智能体编排)
|
||||
- [核心框架](#核心框架)
|
||||
- [多智能体编排框架](#多智能体编排框架)
|
||||
- [群体智能框架](#群体智能框架)
|
||||
- [基于图的编排框架](#基于图的编排框架)
|
||||
- [类型安全 & 现代框架](#类型安全--现代框架)
|
||||
- [TypeScript & Node.js 框架](#typescript--nodejs-框架)
|
||||
- [Go 框架](#go-框架)
|
||||
- [Swift、Ruby 和 Elixir 框架](#swiftruby-和-elixir-框架)
|
||||
- [PHP 和 Java 框架](#php-和-java-框架)
|
||||
- [智能体通信协议](#智能体通信协议)
|
||||
- [Agent-to-Agent (A2A)](#agent-to-agent-a2a)
|
||||
- [Model Context Protocol (MCP)](#model-context-protocol-mcp)
|
||||
- [编排模式](#编排模式)
|
||||
- [层级式](#层级式)
|
||||
- [顺序式](#顺序式)
|
||||
- [并行/并发式](#并行并发式)
|
||||
- [混合专家模式 (MoA)](#混合专家模式-moa)
|
||||
- [智能体委员会](#智能体委员会)
|
||||
- [通信与协调](#通信与协调)
|
||||
- [研究论文](#研究论文)
|
||||
- [工具与基础设施](#工具与基础设施)
|
||||
- [学习资源](#学习资源)
|
||||
- [相关 Awesome 列表](#相关-awesome-列表)
|
||||
|
||||
---
|
||||
|
||||
## 什么是智能体编排？
|
||||
|
||||
**智能体编排**（Agent Orchestration）是协调多个 AI 智能体（LLM智能体、自主智能体）共同完成复杂任务的实践。它能够实现：
|
||||
|
||||
- **可扩展性**：通过将任务分配给专业智能体来处理复杂工作流
|
||||
- **专业化**：每个智能体可以专注于特定领域或任务
|
||||
- **鲁棒性**：通过智能体冗余和恢复实现容错
|
||||
- **涌现能力**：通过简单的智能体交互产生复杂行为
|
||||
- **实际应用**：从软件开发到研究自动化
|
||||
|
||||
### 相关概念
|
||||
|
||||
- **多智能体系统 (MAS)**：由多个交互式智能体组成的系统
|
||||
- **群体智能**：从简单智能体交互中涌现的集体行为
|
||||
- **智能体AI (Agentic AI)**：能够自主规划和执行复杂任务的 AI 系统
|
||||
- **智能体通信协议**：如 A2A (Agent-to-Agent) 和 MCP (Model Context Protocol) 等智能体互操作标准
|
||||
|
||||
---
|
||||
|
||||
## 核心框架
|
||||
|
||||
### 多智能体编排框架
|
||||
|
||||
| 项目 | 描述 | Stars |
|
||||
|------|------|-------|
|
||||
| [AutoGen](https://github.com/microsoft/autogen) | 微软的多智能体会话系统框架 |  |
|
||||
| [CrewAI](https://github.com/crewAIInc/crewAI) | 基于角色的多智能体框架，构建 AI 团队 |  |
|
||||
| [MetaGPT](https://github.com/geekan/MetaGPT) | 软件开发多智能体框架，支持角色分配 |  |
|
||||
| [agentUniverse](https://github.com/agentuniverse-ai/agentUniverse) | 企业级多智能体框架 |  |
|
||||
| [Langroid](https://github.com/langroid/langroid) | 基于消息传递的多智能体编程框架 |  |
|
||||
| [Phidata](https://github.com/agno-agi/phidata) | 多模态智能体框架，内置工具和 UI |  |
|
||||
| [OpenAgents](https://github.com/OpenAgentsInc/openagents) | 开放平台，用于部署语言智能体 |  |
|
||||
| [AgentVerse](https://github.com/OpenBML/AgentVerse) | 在模拟环境中运行多个 LLM 智能体的平台 |  |
|
||||
| [agency-swarm](https://github.com/VRSEN/agency-swarm) | 可靠的多智能体编排框架 |  |
|
||||
| [OpenAI Swarm](https://github.com/openai/swarm) | 教育级轻量级多智能体编排框架 |  |
|
||||
| [Microsoft Agent Framework](https://github.com/microsoft/agent-framework) | 微软构建、编排和部署 AI 智能体的框架 |  |
|
||||
| [OpenAI Agents Python](https://github.com/openai/openai-agents-python) | 轻量级强大的多智能体工作流框架 |  |
|
||||
| [Haystack](https://github.com/deepset-ai/haystack) | 开源 AI 编排框架，用于构建上下文工程化的 LLM 应用 |  |
|
||||
| [LLMStack](https://github.com/trypromptly/LLMStack) | 无代码多智能体框架，构建 LLM 智能体和工作流 |  |
|
||||
| [LazyLLM](https://github.com/LazyAGI/LazyLLM) | 构建多智能体 LLM 应用程序最简单的方式 |  |
|
||||
| [MindSearch](https://github.com/InternLM/MindSearch) | 基于 LLM 的多智能体网络搜索引擎框架 |  |
|
||||
| [TradingAgents](https://github.com/TauricResearch/TradingAgents) | 多智能体 LLM 金融交易框架 |  |
|
||||
| [Shannon](https://github.com/Kocoro-lab/Shannon) | 生产级多智能体编排框架 |  |
|
||||
| [Solace Agent Mesh](https://github.com/SolaceLabs/solace-agent-mesh) | 事件驱动的多智能体 AI 系统框架 |  |
|
||||
| [Pipelex](https://github.com/Pipelex/pipelex) | 可组合 AI 工作流的声明式语言 |  |
|
||||
| [CodeFuse-muAgent](https://github.com/codefuse-ai/CodeFuse-muAgent) | 知识图谱驱动的创新智能体框架 |  |
|
||||
| [L2MAC](https://github.com/samholt/L2MAC) | LLM 自动计算机框架 |  |
|
||||
| [Patchwork](https://github.com/patched-codes/patchwork) | 企业级工作流自动化的智能体 AI 框架 |  |
|
||||
|
||||
### 群体智能框架
|
||||
|
||||
| 项目 | 描述 | Stars |
|
||||
|------|------|-------|
|
||||
| [Swarms](https://github.com/kyegomez/swarms) | 生产级多智能体基础设施平台 |  |
|
||||
| [Swarm](https://github.com/autoreason/swarm) | 轻量级多智能体编排框架 |  |
|
||||
| [Swarm-Tools](https://github.com/FelipeDaza7/swarm-tools) | 协调具有学习能力的 AI 智能体的工具 |  |
|
||||
| [Ruflo](https://github.com/ruvnet/ruflo) | Claude 主导的智能体编排平台 |  |
|
||||
| [swarms-rs](https://github.com/The-Swarm-Corporation/swarms-rs) | Rust 版企业级多智能体编排 |  |
|
||||
| [Lux](https://github.com/Spectral-Finance/lux) | 多智能体群体智能开源框架 |  |
|
||||
| [AutoRT](https://github.com/kyegomez/AutoRT) | 机器人智能体大规模编排的基础模型 |  |
|
||||
|
||||
### 基于图的编排框架
|
||||
|
||||
| 项目 | 描述 | Stars |
|
||||
|------|------|-------|
|
||||
| [LangGraph](https://github.com/langchain-ai/langgraph) | 使用 LangChain 构建有状态的多智能体应用 |  |
|
||||
| [LangChain](https://github.com/langchain-ai/langchain) | 模块化 LLM 应用链接框架 |  |
|
||||
| [LlamaIndex](https://github.com/jerryjliu/llama_index) | 以数据为中心的 RAG 智能体框架 |  |
|
||||
| [Semantic Kernel](https://github.com/microsoft/semantic-kernel) | 微软面向 .NET/Python 的编排 SDK |  |
|
||||
|
||||
### 轻量级和专用智能体
|
||||
|
||||
| 项目 | 描述 | Stars |
|
||||
|------|------|-------|
|
||||
| [SmolAgents](https://github.com/smol-ai/smolagents) | 极简代码编写智能体库 |  |
|
||||
| [Auto-GPT](https://github.com/Significant-Gravitas/Auto-GPT) | 开创性的自主智能体 |  |
|
||||
| [BabyAGI](https://github.com/yoheinakajima/babyagi) | 简单的自主任务管理 |  |
|
||||
| [GPT-Engineer](https://github.com/AntonOsika/gpt-engineer) | 根据规范生成完整项目的智能体 |  |
|
||||
| [BMTools](https://github.com/OpenBMB/BMTools) | 智能体工具/API 注册表 |  |
|
||||
|
||||
### 类型安全 & 现代框架
|
||||
|
||||
| 项目 | 描述 | Stars |
|
||||
|------|------|-------|
|
||||
| [PydanticAI](https://github.com/pydantic/pydantic-ai) | 使用 Pydantic 模型的类型安全智能体框架 |  |
|
||||
| [DeepAgents](https://github.com/vstorm-co/pydantic-deepagents) | 基于 PydanticAI 的生产级自主智能体 |  |
|
||||
| [Subagents](https://github.com/vstorm-co/subagents-pydantic-ai) | PydanticAI 的子智能体委托框架 |  |
|
||||
| [Flux0](https://github.com/flux0-ai/flux0) | 带会话管理的多智能体部署框架 |  |
|
||||
| [Mamba Agents](https://github.com/sequenzia/mamba-agents) | 简单可扩展的 AI 智能体框架 |  |
|
||||
|
||||
### TypeScript & Node.js 框架
|
||||
|
||||
| 项目 | 描述 | Stars |
|
||||
|------|------|-------|
|
||||
| [openai-swarm-node](https://github.com/youseai/openai-swarm-node) | OpenAI Swarm 的 Node.js 实现 |  |
|
||||
| [Network-AI](https://github.com/jovanSAPFIONEER/Network-AI) | TypeScript/Node 多智能体编排器，带共享状态 |  |
|
||||
| [megazord](https://github.com/Sh3rd3n/megazord) | Claude Code 多智能体协作框架 |  |
|
||||
| [Voltmachines](https://github.com/ssdeanx/Voltmachines) | 基于 VoltAgent 的多智能体框架 |  |
|
||||
| [agent-swarm-kit](https://github.com/tripolskypetr/agent-swarm-kit) | TypeScript 库，用于构建框架无关的多智能体系统 |  |
|
||||
| [ai-orchestra](https://github.com/langtail/ai-orchestra) | 轻量级 AI 智能体编排，围绕 Vercel streamText |  |
|
||||
|
||||
### Go 框架
|
||||
|
||||
| 项目 | 描述 | Stars |
|
||||
|------|------|-------|
|
||||
| [openai-agents-go](https://github.com/nlpodyssey/openai-agents-go) | Go 语言轻量级强大的多智能体工作流框架 |  |
|
||||
| [swarm-go](https://github.com/feiskyer/swarm-go) | Go 语言的轻量级多智能体编排 |  |
|
||||
| [rs-graph-llm](https://github.com/a-agmon/rs-graph-llm) | Rust 高性能交互式多智能体工作流框架 |  |
|
||||
|
||||
### Swift、Ruby 和 Elixir 框架
|
||||
|
||||
| 项目 | 描述 | Stars |
|
||||
|------|------|-------|
|
||||
| [SwiftSwarm](https://github.com/jamesrochabrun/SwiftSwarm) | Swift 轻量级多智能体编排框架 |  |
|
||||
| [Swarm](https://github.com/christopherkarani/Swarm) | Swift 6.2 轻量级智能体编排框架 |  |
|
||||
| [ruby-openai-swarm](https://github.com/graysonchen/ruby-openai-swarm) | Ruby 版多智能体编排框架 |  |
|
||||
| [swarm_ex](https://github.com/nrrso/swarm_ex) | Elixir 轻量级 AI 智能体编排库 |  |
|
||||
| [shifts](https://github.com/aaronrussell/shifts) | Elixir 组合自主 AI 智能体工作流的框架 |  |
|
||||
|
||||
### PHP 和 Java 框架
|
||||
|
||||
| 项目 | 描述 | Stars |
|
||||
|------|------|-------|
|
||||
| [neuron-ai](https://github.com/neuron-core/neuron-ai) | PHP 智能体框架，用于构建生产级 AI 应用 |  |
|
||||
| [swarm-ai](https://github.com/intelliswarm-ai/swarm-ai) | Spring AI 的 Java 多智能体编排 |  |
|
||||
|
||||
### 工作流 & 管道框架
|
||||
|
||||
| 项目 | 描述 | Stars |
|
||||
|------|------|-------|
|
||||
| [circuit](https://github.com/smogili1/circuit) | 拖放式智能体工作流构建器 |  |
|
||||
| [routilux](https://github.com/lzjever/routilux) | Python 事件驱动工作流编排 |  |
|
||||
| [CraftFlow](https://github.com/scholarlords/CraftFlow) | RAG 和多智能体协作的工作流编排框架 |  |
|
||||
| [sentinels](https://github.com/garyblankenship/sentinels) | Laravel 智能体任务编排包 |  |
|
||||
| [titan-orchestrator](https://github.com/ramn51/titan-orchestrator) | 连接 DevOps 和 Agentic AI 的分布式编排器 |  |
|
||||
|
||||
---
|
||||
|
||||
## 智能体通信协议
|
||||
|
||||
### Agent-to-Agent (A2A)
|
||||
|
||||
Google 的 A2A 协议支持跨框架和供应商的自主智能体之间的安全、可互操作的通信。
|
||||
|
||||
| 项目 | 描述 | Stars |
|
||||
|------|------|-------|
|
||||
| [awesome-a2a-agents](https://github.com/isekOS/awesome-a2a-agents) | A2A 工具和框架精选列表 |  |
|
||||
| [python-a2a](https://github.com/themanojdesai/python-a2a) | A2A 协议的 Python 实现 |  |
|
||||
| [a2a-net](https://github.com/neuroglia-io/a2a-net) | A2A 协议的 .NET 实现 |  |
|
||||
| [a2a-langgraph](https://github.com/ruska-ai/a2a-langgraph) | 基于 LangGraph 的 A2A 协议 |  |
|
||||
| [nestjs-a2a](https://github.com/thestupd/nestjs-a2a) | A2A 协议的 NestJS 库 |  |
|
||||
| [a2a4j](https://github.com/PheonixHkbxoic/a2a4j) | A2A 协议的 Java 实现 |  |
|
||||
| [A2A-MCP-Server](https://github.com/GongRzhe/A2A-MCP-Server) | MCP 和 A2A 协议之间的桥梁 |  |
|
||||
| [a2a-x402](https://github.com/google-agentic-commerce/a2a-x402) | 支持加密货币支付的 A2A 协议 |  |
|
||||
| [Routa](https://github.com/phodal/routa) | 通过 MCP/A2A 的多智能体协作平台 |  |
|
||||
| [mangaba_ai](https://github.com/Mangaba-ai/mangaba_ai) | 极简 AI 智能体，支持 A2A 和 MCP |  |
|
||||
|
||||
### Model Context Protocol (MCP)
|
||||
|
||||
一种开放协议，使 AI 模型与外部工具/数据源的连接更加无缝。MCP 正在成为工具增强型 LLM 应用的标准。
|
||||
|
||||
#### 官方资源
|
||||
|
||||
| 项目 | 描述 | Stars |
|
||||
|------|------|-------|
|
||||
| [awesome-mcp-servers](https://github.com/wong2/awesome-mcp-servers) | 最流行的 MCP 服务器精选列表 |  |
|
||||
| [modelcontextprotocol/registry](https://github.com/modelcontextprotocol/registry) | 社区驱动的 MCP 服务器注册表 |  |
|
||||
| [microsoft/mcp](https://github.com/microsoft/mcp) | 微软官方 MCP 服务器目录 |  |
|
||||
| [modelcontextprotocol/swift-sdk](https://github.com/modelcontextprotocol/swift-sdk) | 官方 Swift SDK |  |
|
||||
|
||||
#### 数据库与向量存储
|
||||
|
||||
| 项目 | 描述 | Stars |
|
||||
|------|------|-------|
|
||||
| [mcp-server-qdrant](https://github.com/qdrant/mcp-server-qdrant) | 官方 Qdrant 向量数据库服务器 |  |
|
||||
| [mcp-server-milvus](https://github.com/zilliztech/mcp-server-milvus) | Milvus 向量数据库服务器 |  |
|
||||
| [mcp-neo4j](https://github.com/neo4j-contrib/mcp-neo4j) | Neo4j 图数据库服务器 |  |
|
||||
| [mongodb-mcp-server](https://github.com/mongodb-js/mcp-mcp-server) | MongoDB 数据库服务器 |  |
|
||||
| [mysql_mcp_server](https://github.com/designcomputer/mysql_mcp_server) | MySQL 数据库服务器 |  |
|
||||
| [MariaDB/mcp](https://github.com/MariaDB/mcp) | MariaDB 数据库服务器 |  |
|
||||
|
||||
#### 浏览器与自动化
|
||||
|
||||
| 项目 | 描述 | Stars |
|
||||
|------|------|-------|
|
||||
| [mcp-playwright](https://github.com/executeautomation/mcp-playwright) | Playwright 浏览器自动化 |  |
|
||||
| [mcp-chrome](https://github.com/hangwin/mcp-chrome) | Chrome 扩展用于浏览器控制 |  |
|
||||
| [mobile-mcp](https://github.com/mobile-next/mobile-mcp) | iOS/Android 自动化 |  |
|
||||
|
||||
#### 开发与工具
|
||||
|
||||
| 项目 | 描述 | Stars |
|
||||
|------|------|-------|
|
||||
| [mcp-filesystem-server](https://github.com/mark3labs/mcp-filesystem-server) | 文件系统操作 |  |
|
||||
| [jupyter-mcp-server](https://github.com/datalayer/jupyter-mcp-server) | Jupyter notebook 集成 |  |
|
||||
| [fetch-mcp](https://github.com/zcaceres/fetch-mcp) | HTTP 获取 |  |
|
||||
| [mcp-graphql](https://github.com/blurrah/mcp-graphql) | GraphQL API 集成 |  |
|
||||
| [mcpadapt](https://github.com/grll/mcpadapt) | 连接 650+ MCP 服务器到任何框架 |  |
|
||||
|
||||
#### 专业服务器
|
||||
|
||||
| 项目 | 描述 | Stars |
|
||||
|------|------|-------|
|
||||
| [arxiv-mcp-server](https://github.com/blazickjp/arxiv-mcp-server) | arXiv 论文搜索 |  |
|
||||
| [mcp-youtube](https://github.com/anaisbetts/mcp-youtube) | YouTube 视频分析 |  |
|
||||
| [excel-mcp-server](https://github.com/haris-musa/excel-mcp-server) | Excel 文件操作 |  |
|
||||
| [drawio-mcp-server](https://github.com/lgazo/drawio-mcp-server) | 图表创建 |  |
|
||||
| [freecad-mcp](https://github.com/neka-nat/freecad-mcp) | FreeCAD 3D 建模 |  |
|
||||
| [ida-mcp-server](https://github.com/MxIris-Reverse-Engineering/ida-mcp-server) | IDA 逆向工程 |  |
|
||||
| [kubernetes-mcp-server](https://github.com/containers/kubernetes-mcp-server) | Kubernetes 管理 |  |
|
||||
|
||||
---
|
||||
|
||||
## 编排模式
|
||||
|
||||
### 层级式
|
||||
|
||||
在层级式编排中，主管智能体将任务委托给专业子智能体。这种模式适用于需要领域专业知识的复杂任务。
|
||||
|
||||
- **用例**：软件开发团队、研究自动化、企业工作流
|
||||
- **框架**：AutoGen, CrewAI, LangGraph
|
||||
|
||||
### 顺序式
|
||||
|
||||
顺序工作流通过管道处理任务，每个智能体完成其步骤后再传递给下一个。
|
||||
|
||||
- **用例**：文档处理、数据转换管道
|
||||
- **框架**：LangChain, LangGraph
|
||||
|
||||
### 并行/并发式
|
||||
|
||||
多个智能体同时处理独立子任务，之后汇总结果。
|
||||
|
||||
- **用例**：研究收集、并行代码生成
|
||||
- **框架**：Swarms, AutoGen
|
||||
|
||||
### 混合专家模式 (MoA)
|
||||
|
||||
多个专家智能体贡献各自视角，通过迭代细化合并输出。
|
||||
|
||||
- **论文**：[混合专家模式论文](https://arxiv.org/abs/2406.04692)
|
||||
- **实现**：Swarms MoA, LangGraph
|
||||
|
||||
### 智能体委员会
|
||||
|
||||
多个智能体讨论和辩论以达成共识。
|
||||
|
||||
- **用例**：决策、复杂分析、创意写作
|
||||
- **框架**：AutoGen GroupChat
|
||||
|
||||
---
|
||||
|
||||
## 通信与协调
|
||||
|
||||
### 消息传递
|
||||
|
||||
- **Langroid**：基于消息传递的多智能体框架
|
||||
- **AutoGen**：会话消息传递
|
||||
|
||||
### 共享状态/黑板系统
|
||||
|
||||
- **Swarm Matcher**：智能体匹配和选择
|
||||
- **Message Pool**：共享通信系统
|
||||
|
||||
### 任务分配
|
||||
|
||||
- **Auto Swarm**：具有自动任务分配的自组织群体
|
||||
- **Agent Delegation**：任务委托和管理
|
||||
|
||||
---
|
||||
|
||||
## 研究论文
|
||||
|
||||
### 多智能体协作
|
||||
|
||||
| 论文 | 描述 | 年份 |
|
||||
|------|------|------|
|
||||
| [CAMEL: 沟通智能体探索思维](https://arxiv.org/abs/2303.17760) | 通过对话进行角色扮演的智能体协作 | 2023 |
|
||||
| [MetaGPT: 多智能体协作框架](https://arxiv.org/abs/2308.07370) | 软件开发中的智能体角色（产品经理、工程师、测试） | 2023 |
|
||||
| [AgentVerse: 灵活的多智能体平台](https://arxiv.org/abs/2308.07429) | 模拟环境中的协作问题解决 | 2023 |
|
||||
| [生成式智能体](https://arxiv.org/abs/2304.03442) | 虚拟世界中具有长期记忆的交互式智能体 | 2023 |
|
||||
| [混合专家模式](https://arxiv.org/abs/2406.04692) | 并行处理与迭代细化 | 2024 |
|
||||
|
||||
### 编排与规划
|
||||
|
||||
| 论文 | 描述 | 年份 |
|
||||
|------|------|------|
|
||||
| [MALT: 多智能体 LLM 编排](https://arxiv.org/abs/2412.01928) | 创建者-验证者-优化器模式的结构化对话 | 2024 |
|
||||
| [智能体即评判者](https://arxiv.org/abs/2410.10934) | 智能体评估其他智能体 | 2024 |
|
||||
| [ReAct: 推理与行动的协同](https://arxiv.org/abs/2210.03629) | 交错思考和行动 | 2022 |
|
||||
| [Chain-of-Thought Prompting](https://arxiv.org/abs/2201.11903) | LLM 提示中的逐步推理 | 2022 |
|
||||
| [Tree-of-Thoughts](https://arxiv.org/abs/2305.10601) | 复杂问题的多推理分支 | 2023 |
|
||||
| [Reflexion](https://arxiv.org/abs/2309.00668) | 长期任务的自我批评循环 | 2023 |
|
||||
| [HuggingGPT](https://arxiv.org/abs/2303.17580) | LLM 控制器编排专家模型 | 2023 |
|
||||
| [Toolformer](https://arxiv.org/abs/2302.04761) | API 调用的自监督微调 | 2023 |
|
||||
|
||||
### 记忆与长期自主
|
||||
|
||||
| 论文 | 描述 | 年份 |
|
||||
|------|------|------|
|
||||
| [生成式智能体](https://arxiv.org/abs/2304.03442) | 虚拟世界中模拟人类的交互式智能体 | 2023 |
|
||||
| [LLM 智能体记忆机制综述](https://arxiv.org/abs/2402.13736) | 记忆架构和检索概述 | 2024 |
|
||||
|
||||
### 评估
|
||||
|
||||
| 基准 | 描述 |
|
||||
|------|------|
|
||||
| [AgentBench](https://arxiv.org/abs/2308.07258) | 在不同环境中评估智能体 |
|
||||
| [多智能体评估](https://arxiv.org/abs/2312.04323) | 智能体评估指标综述 |
|
||||
|
||||
---
|
||||
|
||||
## 工具与基础设施
|
||||
|
||||
### 可观测性与追踪
|
||||
|
||||
| 工具 | 描述 |
|
||||
|------|------|
|
||||
| [Langfuse](https://github.com/langfuse/langfuse) | 开源 LLM 工程平台 |
|
||||
| [Phoenix](https://github.com/Arize-ai/phoenix) | 开源 LLM 追踪 |
|
||||
| [Opik](https://github.com/comet-ml/opik) | 开源 LLM 评估 |
|
||||
| [Agenta](https://github.com/agenta-ai/agenta) | LLMOps 原型开发平台 |
|
||||
| [Logfire](https://github.com/pydantic/logfire) | Pydantic 的开源可观测性工具 |
|
||||
| [Langtrace](https://github.com/Scale3-Labs/langtrace) | AI 智能体的开源可观测性 |
|
||||
| [Okahu Monocle](https://github.com/OkahuAI/monocle) | Linux 基金会的追踪框架 |
|
||||
|
||||
### 记忆与知识
|
||||
|
||||
| 工具 | 描述 |
|
||||
|------|------|
|
||||
| [Memori](https://github.com/MemoriLabs/Memori) | LLM 和智能体的 SQL 原生记忆层 |
|
||||
| [Letta](https://github.com/letta/letta) | AI 智能体的记忆层 |
|
||||
|
||||
### 部署与基础设施
|
||||
|
||||
| 工具 | 描述 |
|
||||
|------|------|
|
||||
| [OpenAgents](https://github.com/OpenAgentsInc/openagents) | 部署语言智能体的平台 |
|
||||
| [FastAPI Agents](https://github.com/blairhudson/fastapi-agents) | AI 智能体框架的 FastAPI 扩展 |
|
||||
| [AgentAPIProduction](https://github.com/The-Swarm-Corporation/AgentAPIProduction) | 生产级智能体 API 系统 |
|
||||
| [AgentOS](https://github.com/The-Swarm-Corporation/AgentOS) | AI 智能体操作系统 |
|
||||
|
||||
### 评估与基准测试
|
||||
|
||||
| 工具 | 描述 |
|
||||
|------|------|
|
||||
| [AgentBench](https://github.com/AgentBench/AgentBench) | 在不同环境中评估智能体 |
|
||||
| [swarms-evals](https://github.com/The-Swarm-Corporation/swarms-evals) | 群体系统的评估框架 |
|
||||
| [RAGAS](https://github.com/explodinggradients/ragas) | RAG 系统的评估框架 |
|
||||
|
||||
---
|
||||
|
||||
## 学习资源
|
||||
|
||||
### 教程与指南
|
||||
|
||||
- [LLM 驱动的自主智能体](https://lilianweng.github.io/posts/2023-06-15-agent/) - Lilian Weng 的综合指南
|
||||
- [提示工程指南：LLM 智能体](https://github.com/dair-ai/Prompt-Engineering-Guide#llm-agents) - 全面的智能体构建指南
|
||||
- [使用 LangChain 构建自主智能体](https://www.youtube.com/watch?v=example) - 视频教程
|
||||
- [企业级智能体课程](https://github.com/The-Swarm-Corporation/Enterprise-Grade-Agents-Course) - 综合课程
|
||||
|
||||
### 文档
|
||||
|
||||
- [AutoGen 文档](https://microsoft.github.io/autogen/) - 微软 AutoGen 文档
|
||||
- [CrewAI 文档](https://docs.crewai.com/) - CrewAI 官方文档
|
||||
- [LangGraph 文档](https://langchain-ai.github.io/langgraph/) - LangGraph 文档
|
||||
- [Swarms 文档](https://docs.swarms.world/) - Swarms 框架文档
|
||||
|
||||
---
|
||||
|
||||
## 相关 Awesome 列表
|
||||
|
||||
- [awesome-ai-agents](https://github.com/heurema/awesome-ai-agents) - 综合 AI 智能体资源
|
||||
- [awesome-LangGraph](https://github.com/von-development/awesome-LangGraph) - LangChain/LangGraph 生态系统
|
||||
- [awesome-multi-agent-papers](https://github.com/kyegomez/awesome-multi-agent-papers) - 多智能体研究论文
|
||||
- [awesome-Swarms-List](https://github.com/The-Swarm-Corporation/Awesome-Swarms-List) - Swarms 框架资源
|
||||
- [awesome-a2a-agents](https://github.com/isekOS/awesome-a2a-agents) - A2A 协议生态系统
|
||||
- [awesome-agent-protocols](https://github.com/shanjai-raj/awesome-agent-protocols) - 智能体通信协议
|
||||
- [awesome-agents](https://github.com/l-aime/awesome-agents) - 前沿 AI 智能体项目
|
||||
- [awesome-agentic-ai](https://github.com/mlnjsh/awesome-agentic-ai) - 智能体 AI 框架和论文
|
||||
|
||||
---
|
||||
|
||||
## 贡献
|
||||
|
||||
欢迎贡献！请先阅读[贡献指南](CONTRIBUTING.md)。
|
||||
|
||||
1. Fork 本仓库
|
||||
2. 在相应部分添加您的资源
|
||||
3. 确保格式一致
|
||||
4. 提交 Pull Request
|
||||
|
||||
---
|
||||
|
||||
## SEO 关键词
|
||||
|
||||
> 热门搜索词：AI智能体框架、多智能体系统、智能体编排、群体智能、LLM智能体、Claude AI、OpenAI智能体、CrewAI、AutoGen、MetaGPT、LangGraph、智能体通信协议、A2A协议、MCP协议、智能体AI、自主智能体、智能体工作流、AI自动化
|
||||
|
||||
### 热门框架
|
||||
|
||||
- [AutoGen](https://github.com/microsoft/autogen) - 微软多智能体框架
|
||||
- [CrewAI](https://github.com/crewAIInc/crewAI) - 基于角色的多智能体框架
|
||||
- [MetaGPT](https://github.com/geekan/MetaGPT) - 软件开发多智能体框架
|
||||
- [LangGraph](https://github.com/langchain-ai/langgraph) - 基于状态的智能体工作流
|
||||
- [Swarms](https://github.com/kyegomez/swarms) - 生产级多智能体平台
|
||||
|
||||
### 热门协议
|
||||
|
||||
- [A2A 协议](https://google.github.io/A2A/) - 智能体间通信协议
|
||||
- [MCP](https://modelcontextprotocol.io/) - 模型上下文协议
|
||||
|
||||
---
|
||||
|
||||
## 许可证
|
||||
|
||||
[](LICENSE)
|
||||
|
||||
根据法律允许的范围，[您的名字](https://github.com/yourusername) 已放弃本作品的所有版权和相关权利。
|
||||
|
||||
---
|
||||
|
||||
*如果觉得有用，请给本仓库点个 Star！*
|
||||
Loading…
Reference in New Issue