[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-kyegomez--swarms":3,"tool-kyegomez--swarms":64},[4,17,27,35,43,56],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":16},3808,"stable-diffusion-webui","AUTOMATIC1111\u002Fstable-diffusion-webui","stable-diffusion-webui 是一个基于 Gradio 构建的网页版操作界面，旨在让用户能够轻松地在本地运行和使用强大的 Stable Diffusion 图像生成模型。它解决了原始模型依赖命令行、操作门槛高且功能分散的痛点，将复杂的 AI 绘图流程整合进一个直观易用的图形化平台。\n\n无论是希望快速上手的普通创作者、需要精细控制画面细节的设计师，还是想要深入探索模型潜力的开发者与研究人员，都能从中获益。其核心亮点在于极高的功能丰富度：不仅支持文生图、图生图、局部重绘（Inpainting）和外绘（Outpainting）等基础模式，还独创了注意力机制调整、提示词矩阵、负向提示词以及“高清修复”等高级功能。此外，它内置了 GFPGAN 和 CodeFormer 等人脸修复工具，支持多种神经网络放大算法，并允许用户通过插件系统无限扩展能力。即使是显存有限的设备，stable-diffusion-webui 也提供了相应的优化选项，让高质量的 AI 艺术创作变得触手可及。",162132,3,"2026-04-05T11:01:52",[13,14,15],"开发框架","图像","Agent","ready",{"id":18,"name":19,"github_repo":20,"description_zh":21,"stars":22,"difficulty_score":23,"last_commit_at":24,"category_tags":25,"status":16},1381,"everything-claude-code","affaan-m\u002Feverything-claude-code","everything-claude-code 是一套专为 AI 编程助手（如 Claude Code、Codex、Cursor 等）打造的高性能优化系统。它不仅仅是一组配置文件，而是一个经过长期实战打磨的完整框架，旨在解决 AI 代理在实际开发中面临的效率低下、记忆丢失、安全隐患及缺乏持续学习能力等核心痛点。\n\n通过引入技能模块化、直觉增强、记忆持久化机制以及内置的安全扫描功能，everything-claude-code 能显著提升 AI 在复杂任务中的表现，帮助开发者构建更稳定、更智能的生产级 AI 代理。其独特的“研究优先”开发理念和针对 Token 消耗的优化策略，使得模型响应更快、成本更低，同时有效防御潜在的攻击向量。\n\n这套工具特别适合软件开发者、AI 研究人员以及希望深度定制 AI 工作流的技术团队使用。无论您是在构建大型代码库，还是需要 AI 协助进行安全审计与自动化测试，everything-claude-code 都能提供强大的底层支持。作为一个曾荣获 Anthropic 黑客大奖的开源项目，它融合了多语言支持与丰富的实战钩子（hooks），让 AI 真正成长为懂上",138956,2,"2026-04-05T11:33:21",[13,15,26],"语言模型",{"id":28,"name":29,"github_repo":30,"description_zh":31,"stars":32,"difficulty_score":23,"last_commit_at":33,"category_tags":34,"status":16},2271,"ComfyUI","Comfy-Org\u002FComfyUI","ComfyUI 是一款功能强大且高度模块化的视觉 AI 引擎，专为设计和执行复杂的 Stable Diffusion 图像生成流程而打造。它摒弃了传统的代码编写模式，采用直观的节点式流程图界面，让用户通过连接不同的功能模块即可构建个性化的生成管线。\n\n这一设计巧妙解决了高级 AI 绘图工作流配置复杂、灵活性不足的痛点。用户无需具备编程背景，也能自由组合模型、调整参数并实时预览效果，轻松实现从基础文生图到多步骤高清修复等各类复杂任务。ComfyUI 拥有极佳的兼容性，不仅支持 Windows、macOS 和 Linux 全平台，还广泛适配 NVIDIA、AMD、Intel 及苹果 Silicon 等多种硬件架构，并率先支持 SDXL、Flux、SD3 等前沿模型。\n\n无论是希望深入探索算法潜力的研究人员和开发者，还是追求极致创作自由度的设计师与资深 AI 绘画爱好者，ComfyUI 都能提供强大的支持。其独特的模块化架构允许社区不断扩展新功能，使其成为当前最灵活、生态最丰富的开源扩散模型工具之一，帮助用户将创意高效转化为现实。",107662,"2026-04-03T11:11:01",[13,14,15],{"id":36,"name":37,"github_repo":38,"description_zh":39,"stars":40,"difficulty_score":23,"last_commit_at":41,"category_tags":42,"status":16},3704,"NextChat","ChatGPTNextWeb\u002FNextChat","NextChat 是一款轻量且极速的 AI 助手，旨在为用户提供流畅、跨平台的大模型交互体验。它完美解决了用户在多设备间切换时难以保持对话连续性，以及面对众多 AI 模型不知如何统一管理的痛点。无论是日常办公、学习辅助还是创意激发，NextChat 都能让用户随时随地通过网页、iOS、Android、Windows、MacOS 或 Linux 端无缝接入智能服务。\n\n这款工具非常适合普通用户、学生、职场人士以及需要私有化部署的企业团队使用。对于开发者而言，它也提供了便捷的自托管方案，支持一键部署到 Vercel 或 Zeabur 等平台。\n\nNextChat 的核心亮点在于其广泛的模型兼容性，原生支持 Claude、DeepSeek、GPT-4 及 Gemini Pro 等主流大模型，让用户在一个界面即可自由切换不同 AI 能力。此外，它还率先支持 MCP（Model Context Protocol）协议，增强了上下文处理能力。针对企业用户，NextChat 提供专业版解决方案，具备品牌定制、细粒度权限控制、内部知识库整合及安全审计等功能，满足公司对数据隐私和个性化管理的高标准要求。",87618,"2026-04-05T07:20:52",[13,26],{"id":44,"name":45,"github_repo":46,"description_zh":47,"stars":48,"difficulty_score":23,"last_commit_at":49,"category_tags":50,"status":16},2268,"ML-For-Beginners","microsoft\u002FML-For-Beginners","ML-For-Beginners 是由微软推出的一套系统化机器学习入门课程，旨在帮助零基础用户轻松掌握经典机器学习知识。这套课程将学习路径规划为 12 周，包含 26 节精炼课程和 52 道配套测验，内容涵盖从基础概念到实际应用的完整流程，有效解决了初学者面对庞大知识体系时无从下手、缺乏结构化指导的痛点。\n\n无论是希望转型的开发者、需要补充算法背景的研究人员，还是对人工智能充满好奇的普通爱好者，都能从中受益。课程不仅提供了清晰的理论讲解，还强调动手实践，让用户在循序渐进中建立扎实的技能基础。其独特的亮点在于强大的多语言支持，通过自动化机制提供了包括简体中文在内的 50 多种语言版本，极大地降低了全球不同背景用户的学习门槛。此外，项目采用开源协作模式，社区活跃且内容持续更新，确保学习者能获取前沿且准确的技术资讯。如果你正寻找一条清晰、友好且专业的机器学习入门之路，ML-For-Beginners 将是理想的起点。",84991,"2026-04-05T10:45:23",[14,51,52,53,15,54,26,13,55],"数据工具","视频","插件","其他","音频",{"id":57,"name":58,"github_repo":59,"description_zh":60,"stars":61,"difficulty_score":10,"last_commit_at":62,"category_tags":63,"status":16},3128,"ragflow","infiniflow\u002Fragflow","RAGFlow 是一款领先的开源检索增强生成（RAG）引擎，旨在为大语言模型构建更精准、可靠的上下文层。它巧妙地将前沿的 RAG 技术与智能体（Agent）能力相结合，不仅支持从各类文档中高效提取知识，还能让模型基于这些知识进行逻辑推理和任务执行。\n\n在大模型应用中，幻觉问题和知识滞后是常见痛点。RAGFlow 通过深度解析复杂文档结构（如表格、图表及混合排版），显著提升了信息检索的准确度，从而有效减少模型“胡编乱造”的现象，确保回答既有据可依又具备时效性。其内置的智能体机制更进一步，使系统不仅能回答问题，还能自主规划步骤解决复杂问题。\n\n这款工具特别适合开发者、企业技术团队以及 AI 研究人员使用。无论是希望快速搭建私有知识库问答系统，还是致力于探索大模型在垂直领域落地的创新者，都能从中受益。RAGFlow 提供了可视化的工作流编排界面和灵活的 API 接口，既降低了非算法背景用户的上手门槛，也满足了专业开发者对系统深度定制的需求。作为基于 Apache 2.0 协议开源的项目，它正成为连接通用大模型与行业专有知识之间的重要桥梁。",77062,"2026-04-04T04:44:48",[15,14,13,26,54],{"id":65,"github_repo":66,"name":67,"description_en":68,"description_zh":69,"ai_summary_zh":69,"readme_en":70,"readme_zh":71,"quickstart_zh":72,"use_case_zh":73,"hero_image_url":74,"owner_login":75,"owner_name":76,"owner_avatar_url":77,"owner_bio":78,"owner_company":79,"owner_location":80,"owner_email":81,"owner_twitter":82,"owner_website":83,"owner_url":84,"languages":85,"stars":102,"forks":103,"last_commit_at":104,"license":105,"difficulty_score":23,"env_os":106,"env_gpu":107,"env_ram":107,"env_deps":108,"category_tags":111,"github_topics":112,"view_count":130,"oss_zip_url":81,"oss_zip_packed_at":81,"status":16,"created_at":131,"updated_at":132,"faqs":133,"releases":164},160,"kyegomez\u002Fswarms","swarms","The Enterprise-Grade Production-Ready Multi-Agent Orchestration Framework. Website: https:\u002F\u002Fswarms.ai","Swarms 是一个面向企业级应用的多智能体协同框架，专为生产环境设计，帮助开发者高效构建、部署和管理由多个 AI 智能体组成的复杂系统。它解决了传统单智能体在处理复杂任务时能力有限、协作困难的问题，通过支持分层编排、并行流水线、图结构网络等机制，让多个智能体像“蜂群”一样协同完成业务流程自动化、知识推理或决策支持等任务。\n\nSwarms 适合 AI 工程师、系统架构师以及希望将多智能体技术落地到实际业务中的开发团队使用。它兼容 LangChain、AutoGen、CrewAI 等主流框架，支持多种大模型提供商，并提供模块化微服务架构、可观测性工具和自动扩缩容能力，便于与现有企业系统集成。其独特的动态智能体组合、统一注册管理和标准化 API 接口，显著降低了多智能体系统的开发与运维复杂度，在保证高可用的同时提升资源利用效率。","\u003Cdiv align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fswarms.world\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fkyegomez_swarms_readme_1df629ea353f.png\" style=\"margin: 15px; max-width: 350px\" width=\"80%\" alt=\"Logo\">\n  \u003C\u002Fa>\n\u003C\u002Fdiv>\n\u003Cp align=\"center\">\n  \u003Cem>The Enterprise-Grade Production-Ready Multi-Agent Orchestration Framework \u003C\u002Fem>\n\u003C\u002Fp>\n\n\n\u003Cp align=\"center\">\n  \u003C!-- Main Navigation Links -->\n  \u003Ca href=\"https:\u002F\u002Fswarms.ai\">Swarms Website\u003C\u002Fa>\n  \u003Cspan>  •  \u003C\u002Fspan>\n  \u003Ca href=\"https:\u002F\u002Fdocs.swarms.world\">Documentation\u003C\u002Fa>\n  \u003Cspan>  •  \u003C\u002Fspan>\n  \u003Ca href=\"https:\u002F\u002Fswarms.world\">Swarms Marketplace\u003C\u002Fa>\n\u003C\u002Fp>\n\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fpypi.org\u002Fproject\u002Fswarms\u002F\" target=\"_blank\">\n    \u003Cpicture>\n      \u003Csource srcset=\"https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fv\u002Fswarms?style=for-the-badge&color=3670A0\" media=\"(prefers-color-scheme: dark)\">\n      \u003Cimg alt=\"Version\" src=\"https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fv\u002Fswarms?style=for-the-badge&color=3670A0\">\n    \u003C\u002Fpicture>\n  \u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fpypi.org\u002Fproject\u002Fswarms\u002F\" target=\"_blank\">\n    \u003Cpicture>\n      \u003Csource srcset=\"https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fdm\u002Fswarms?style=for-the-badge&color=3670A0\" media=\"(prefers-color-scheme: dark)\">\n      \u003Cimg alt=\"Downloads\" src=\"https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fdm\u002Fswarms?style=for-the-badge&color=3670A0\">\n    \u003C\u002Fpicture>\n  \u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Ftwitter.com\u002Fswarms_corp\u002F\">\n    \u003Cpicture>\n      \u003Csource srcset=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FTwitter-Follow-1DA1F2?style=for-the-badge&logo=twitter&logoColor=white\" media=\"(prefers-color-scheme: dark)\">\n      \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FTwitter-Follow-1DA1F2?style=for-the-badge&logo=twitter&logoColor=white\" alt=\"Twitter\">\n    \u003C\u002Fpicture>\n  \u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fdiscord.gg\u002FEamjgSaEQf\">\n    \u003Cpicture>\n      \u003Csource srcset=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDiscord-Join-5865F2?style=for-the-badge&logo=discord&logoColor=white\" media=\"(prefers-color-scheme: dark)\">\n      \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDiscord-Join-5865F2?style=for-the-badge&logo=discord&logoColor=white\" alt=\"Discord\">\n    \u003C\u002Fpicture>\n  \u003C\u002Fa>\n\u003C\u002Fp>\n\n## Features\n\nSwarms delivers a comprehensive, enterprise-grade multi-agent infrastructure platform designed for production-scale deployments and seamless integration with existing systems. [Learn more about the swarms feature set here](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Ffeatures\u002F)\n\n| Category | Features | Benefits |\n|----------|----------|-----------|\n| **Enterprise Architecture** | • Production-Ready Infrastructure\u003Cbr>• High Availability Systems\u003Cbr>• Modular Microservices Design\u003Cbr>• Comprehensive Observability\u003Cbr>• Backwards Compatibility | • 99.9%+ Uptime Guarantee\u003Cbr>• Reduced Operational Overhead\u003Cbr>• Seamless Legacy Integration\u003Cbr>• Enhanced System Monitoring\u003Cbr>• Risk-Free Migration Path |\n| **Multi-Agent Orchestration** | • Hierarchical Agent Swarms\u003Cbr>• Parallel Processing Pipelines\u003Cbr>• Sequential Workflow Orchestration\u003Cbr>• Graph-Based Agent Networks\u003Cbr>• Dynamic Agent Composition\u003Cbr>• Agent Registry Management | • Complex Business Process Automation\u003Cbr>• Scalable Task Distribution\u003Cbr>• Flexible Workflow Adaptation\u003Cbr>• Optimized Resource Utilization\u003Cbr>• Centralized Agent Governance\u003Cbr>• Enterprise-Grade Agent Lifecycle Management |\n| **Enterprise Integration** | • Multi-Model Provider Support\u003Cbr>• Custom Agent Development Framework\u003Cbr>• Extensive Enterprise Tool Library\u003Cbr>• Multiple Memory Systems\u003Cbr>• Backwards Compatibility with LangChain, AutoGen, CrewAI\u003Cbr>• Standardized API Interfaces | • Vendor-Agnostic Architecture\u003Cbr>• Custom Solution Development\u003Cbr>• Extended Functionality Integration\u003Cbr>• Enhanced Knowledge Management\u003Cbr>• Seamless Framework Migration\u003Cbr>• Reduced Integration Complexity |\n| **Enterprise Scalability** | • Concurrent Multi-Agent Processing\u003Cbr>• Intelligent Resource Management\u003Cbr>• Load Balancing & Auto-Scaling\u003Cbr>• Horizontal Scaling Capabilities\u003Cbr>• Performance Optimization\u003Cbr>• Capacity Planning Tools | • High-Throughput Processing\u003Cbr>• Cost-Effective Resource Utilization\u003Cbr>• Elastic Scaling Based On Demand\u003Cbr>• Linear Performance Scaling\u003Cbr>• Optimized Response Times\u003Cbr>• Predictable Growth Planning |\n| **Developer Experience** | • Intuitive Enterprise API\u003Cbr>• Comprehensive Documentation\u003Cbr>• Active Enterprise Community\u003Cbr>• CLI & SDK Tools\u003Cbr>• IDE Integration Support\u003Cbr>• Code Generation Templates | • Accelerated Development Cycles\u003Cbr>• Reduced Learning Curve\u003Cbr>• Expert Community Support\u003Cbr>• Rapid Deployment Capabilities\u003Cbr>• Enhanced Developer Productivity\u003Cbr>• Standardized Development Patterns |\n\n\n## Supported Protocols & Integrations\n\nSwarms seamlessly integrates with industry-standard protocols and open specifications, unlocking powerful capabilities for tool integration, payment processing, distributed agent orchestration, and model interoperability.\n\n| Protocol | Description | Use Cases | Documentation |\n|----------|-------------|-----------|---------------|\n| **[MCP (Model Context Protocol)](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fmulti_mcp_agent\u002F)** | Standardized protocol for AI agents to interact with external tools and services through MCP servers. Enables dynamic tool discovery and execution. | • Tool integration\u003Cbr>• Multi-server connections\u003Cbr>• External API access\u003Cbr>• Database connectivity | [MCP Integration Guide](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fmulti_mcp_agent\u002F) |\n| **[X402](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fexamples\u002Fx402_payment_integration\u002F)** | Cryptocurrency payment protocol for API endpoints. Enables monetization of agents with pay-per-use models. | • Agent monetization\u003Cbr>• Payment gate protection\u003Cbr>• Crypto payments\u003Cbr>• Pay-per-use services | [X402 Quickstart](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fexamples\u002Fx402_payment_integration\u002F) |\n| **[AOP (Agent Orchestration Protocol)](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fexamples\u002Faop_medical\u002F)** | Framework for deploying and managing agents as distributed services. Enables agent discovery, management, and execution through standardized protocols. | • Distributed agent deployment\u003Cbr>• Agent discovery\u003Cbr>• Service orchestration\u003Cbr>• Scalable multi-agent systems | [AOP Reference](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Faop\u002F) |\n| **[Swarms Marketplace](https:\u002F\u002Fswarms.world)** | Platform for discovering and sharing production-ready prompts, agents, and tools. Enables automatic prompt loading from the marketplace and publishing your own prompts directly from code. | • Prompt discovery and reuse\u003Cbr>• One-line prompt loading\u003Cbr>• Community prompt sharing\u003Cbr>• Prompt monetization | [Marketplace Tutorial](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fmarketplace_prompt_loading\u002F) |\n| **[Open Responses](https:\u002F\u002Fwww.openresponses.org\u002F)** | Open-source specification and ecosystem for multi-provider, interoperable LLM interfaces based on the OpenAI Responses API. Provides a unified schema and tooling for calling language models, streaming results, and composing agentic workflows—independent of provider. | • Unified LLM interfaces\u003Cbr>• Streaming outputs\u003Cbr>• Multi-provider orchestration\u003Cbr>• Interoperable agent workflows | [Open Responses Website](https:\u002F\u002Fwww.openresponses.org\u002F) |\n| **[Agent Skills](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fagents\u002Fagent_skills\u002F)** | Lightweight, markdown-based format for defining modular, reusable agent capabilities introduced by Anthropic. Enables specialization of agents without modifying code by loading skill definitions from simple SKILL.md files. | • Agent specialization\u003Cbr>• Reusable skill libraries\u003Cbr>• Code-free agent customization\u003Cbr>• Claude Code compatibility | [Agent Skills Documentation](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fagents\u002Fagent_skills\u002F) |\n\n\n## Install\n\n### Using pip\n\n```bash\n$ pip3 install -U swarms\n```\n\n### Using uv (Recommended)\n\n[uv](https:\u002F\u002Fgithub.com\u002Fastral-sh\u002Fuv) is a fast Python package installer and resolver, written in Rust.\n\n```bash\n$ uv pip install swarms\n```\n\n### Using poetry\n\n```bash\n$ poetry add swarms\n```\n\n### From source\n\n```bash\n# Clone the repository\n$ git clone https:\u002F\u002Fgithub.com\u002Fkyegomez\u002Fswarms.git\n$ cd swarms\n$ pip install -r requirements.txt\n```\n\n\u003C!-- ### Using Docker\n\nThe easiest way to get started with Swarms is using our pre-built Docker image:\n\n```bash\n# Pull and run the latest image\n$ docker pull kyegomez\u002Fswarms:latest\n$ docker run --rm kyegomez\u002Fswarms:latest python -c \"import swarms; print('Swarms is ready!')\"\n\n# Run interactively for development\n$ docker run -it --rm -v $(pwd):\u002Fapp kyegomez\u002Fswarms:latest bash\n\n# Using docker-compose (recommended for development)\n$ docker-compose up -d\n```\n\nFor more Docker options and advanced usage, see our [Docker documentation](\u002Fscripts\u002Fdocker\u002FDOCKER.md). -->\n\n---\n\n## Environment Configuration\n\n[Learn more about the environment configuration here](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Finstall\u002Fenv\u002F)\n\n```\nOPENAI_API_KEY=\"\"\nWORKSPACE_DIR=\"agent_workspace\"\nANTHROPIC_API_KEY=\"\"\nGROQ_API_KEY=\"\"\n```\n\n\n### Your First Agent\n\nAn **Agent** is the fundamental building block of a swarm—an autonomous entity powered by an LLM + Tools + Memory. [Learn more Here](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fagent\u002F)\n\n```python\nfrom swarms import Agent\n\n# Initialize a new agent\nagent = Agent(\n    model_name=\"gpt-5.4\", # Specify the LLM\n    max_loops=\"auto\",              # Set the number of interactions\n    interactive=True,         # Enable interactive mode for real-time feedback\n)\n\n# Run the agent with a task\nagent.run(\"What are the key benefits of using a multi-agent system?\")\n```\n\n### Your First Swarm: Multi-Agent Collaboration\n\nA **Swarm** consists of multiple agents working together. This simple example creates a two-agent workflow for researching and writing a blog post. [Learn More About SequentialWorkflow](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fsequential_workflow\u002F)\n\n```python\nfrom swarms import Agent, SequentialWorkflow\n\n# Agent 1: The Researcher\nresearcher = Agent(\n    agent_name=\"Researcher\",\n    system_prompt=\"Your job is to research the provided topic and provide a detailed summary.\",\n    model_name=\"gpt-5.4\",\n)\n\n# Agent 2: The Writer\nwriter = Agent(\n    agent_name=\"Writer\",\n    system_prompt=\"Your job is to take the research summary and write a beautiful, engaging blog post about it.\",\n    model_name=\"gpt-5.4\",\n)\n\n# Create a sequential workflow where the researcher's output feeds into the writer's input\nworkflow = SequentialWorkflow(agents=[researcher, writer])\n\n# Run the workflow on a task\nfinal_post = workflow.run(\"The history and future of artificial intelligence\")\nprint(final_post)\n\n```\n\n-----\n\n### AutoSwarmBuilder: Autonomous Agent Generation\n\nThe `AutoSwarmBuilder` automatically generates specialized agents and their workflows based on your task description. Simply describe what you need, and it will create a complete multi-agent system with detailed prompts and optimal agent configurations. [Learn more about AutoSwarmBuilder](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fauto_swarm_builder\u002F)\n\n```python\nfrom swarms.structs.auto_swarm_builder import AutoSwarmBuilder\nimport json\n\n# Initialize the AutoSwarmBuilder\nswarm = AutoSwarmBuilder(\n    name=\"My Swarm\",\n    description=\"A swarm of agents\",\n    verbose=True,\n    max_loops=1,\n    return_agents=True,\n    model_name=\"gpt-5.4\",\n)\n\n# Let the builder automatically create agents and workflows\nresult = swarm.run(\n    task=\"Create an accounting team to analyze crypto transactions, \"\n         \"there must be 5 agents in the team with extremely extensive prompts. \"\n         \"Make the prompts extremely detailed and specific and long and comprehensive. \"\n         \"Make sure to include all the details of the task in the prompts.\"\n)\n\n# The result contains the generated agents and their configurations\nprint(json.dumps(result, indent=4))\n```\n\nThe `AutoSwarmBuilder` provides:\n\n- **Automatic Agent Generation**: Creates specialized agents based on task requirements\n- **Intelligent Prompt Engineering**: Generates comprehensive, detailed prompts for each agent\n- **Optimal Workflow Design**: Determines the best agent interactions and workflow structure\n- **Production-Ready Configurations**: Returns fully configured agents ready for deployment\n- **Flexible Architecture**: Supports various swarm types and agent specializations\n\nThis feature is perfect for rapid prototyping, complex task decomposition, and creating specialized agent teams without manual configuration.\n\n-----\n\n## Available Multi-Agent Architectures\n\n`swarms` provides a variety of powerful, pre-built multi-agent architectures enabling you to orchestrate agents in various ways. Choose the right structure for your specific problem to build efficient and reliable production systems.\n\n| **Architecture** | **Description** | **Best For** |\n|---|---|---|\n| **[SequentialWorkflow](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fsequential_workflow\u002F)** | Agents execute tasks in a linear chain; the output of one agent becomes the input for the next. | Step-by-step processes such as data transformation pipelines and report generation. |\n| **[ConcurrentWorkflow](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fconcurrentworkflow\u002F)** | Agents run tasks simultaneously for maximum efficiency. | High-throughput tasks such as batch processing and parallel data analysis. |\n| **[AgentRearrange](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fagent_rearrange\u002F)** | Dynamically maps complex relationships (e.g., `a -> b, c`) between agents. | Flexible and adaptive workflows, task distribution, and dynamic routing. |\n| **[GraphWorkflow](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fgraph_workflow\u002F)** | Orchestrates agents as nodes in a Directed Acyclic Graph (DAG). | Complex projects with intricate dependencies, such as software builds. |\n| **[MixtureOfAgents (MoA)](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fmoa\u002F)** | Utilizes multiple expert agents in parallel and synthesizes their outputs. | Complex problem-solving and achieving state-of-the-art performance through collaboration. |\n| **[GroupChat](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fgroup_chat\u002F)** | Agents collaborate and make decisions through a conversational interface. | Real-time collaborative decision-making, negotiations, and brainstorming. |\n| **[ForestSwarm](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fforest_swarm\u002F)** | Dynamically selects the most suitable agent or tree of agents for a given task. | Task routing, optimizing for expertise, and complex decision-making trees. |\n| **[HierarchicalSwarm](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fhierarchical_swarm\u002F)** | Orchestrates agents with a director who creates plans and distributes tasks to specialized worker agents. | Complex project management, team coordination, and hierarchical decision-making with feedback loops. |\n| **[HeavySwarm](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fheavy_swarm\u002F)** | Implements a five-phase workflow with specialized agents (Research, Analysis, Alternatives, Verification) for comprehensive task analysis. | Complex research and analysis tasks, financial analysis, strategic planning, and comprehensive reporting. |\n| **[MAKER](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fmaker\u002F)** | Long-horizon tasks decomposed into steps; each step uses first-to-ahead-by-k voting and red-flagging on micro-agent samples (from Meyerson et al., 2025). | Extremely long or fragile pipelines where you want statistical agreement and validation on every atomic step—not a hand-designed multi-agent graph. |\n| **[SwarmRouter](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fswarm_router\u002F)** | A universal orchestrator that provides a single interface to run any type of swarm with dynamic selection. | Simplifying complex workflows, switching between swarm strategies, and unified multi-agent management. |\n\n-----\n\n### SequentialWorkflow\n\nA `SequentialWorkflow` executes tasks in a strict order, forming a pipeline where each agent builds upon the work of the previous one. `SequentialWorkflow` is Ideal for processes that have clear, ordered steps. This ensures that tasks with dependencies are handled correctly.\n\n```python\nfrom swarms import Agent, SequentialWorkflow\n\n# Agent 1: The Researcher\nresearcher = Agent(\n    agent_name=\"Researcher\",\n    system_prompt=\"Your job is to research the provided topic and provide a detailed summary.\",\n    model_name=\"gpt-5.4\",\n)\n\n# Agent 2: The Writer\nwriter = Agent(\n    agent_name=\"Writer\",\n    system_prompt=\"Your job is to take the research summary and write a beautiful, engaging blog post about it.\",\n    model_name=\"gpt-5.4\",\n)\n\n# Create a sequential workflow where the researcher's output feeds into the writer's input\nworkflow = SequentialWorkflow(agents=[researcher, writer])\n\n# Run the workflow on a task\nfinal_post = workflow.run(\"The history and future of artificial intelligence\")\nprint(final_post)\n```\n\n-----\n\n\n### ConcurrentWorkflow\n\nA `ConcurrentWorkflow` runs multiple agents simultaneously, allowing for parallel execution of tasks. This architecture drastically reduces execution time for tasks that can be performed in parallel, making it ideal for high-throughput scenarios where agents work on similar tasks concurrently.\n\n```python\nfrom swarms import Agent, ConcurrentWorkflow\n\n# Create agents for different analysis tasks\nmarket_analyst = Agent(\n    agent_name=\"Market-Analyst\",\n    system_prompt=\"Analyze market trends and provide insights on the given topic.\",\n    model_name=\"gpt-5.4\",\n    max_loops=1,\n)\n\nfinancial_analyst = Agent(\n    agent_name=\"Financial-Analyst\", \n    system_prompt=\"Provide financial analysis and recommendations on the given topic.\",\n    model_name=\"gpt-5.4\",\n    max_loops=1,\n)\n\nrisk_analyst = Agent(\n    agent_name=\"Risk-Analyst\",\n    system_prompt=\"Assess risks and provide risk management strategies for the given topic.\",\n    model_name=\"gpt-5.4\", \n    max_loops=1,\n)\n\n# Create concurrent workflow\nconcurrent_workflow = ConcurrentWorkflow(\n    agents=[market_analyst, financial_analyst, risk_analyst],\n    max_loops=1,\n)\n\n# Run all agents concurrently on the same task\nresults = concurrent_workflow.run(\n    \"Analyze the potential impact of AI technology on the healthcare industry\"\n)\n\nprint(results)\n```\n\n---\n\n### AgentRearrange\n\nInspired by `einsum`, `AgentRearrange` lets you define complex, non-linear relationships between agents using a simple string-based syntax. [Learn more](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fagent_rearrange\u002F). This architecture is perfect for orchestrating dynamic workflows where agents might work in parallel, in sequence, or in any combination you choose.\n\n```python\nfrom swarms import Agent, AgentRearrange\n\n# Define agents\nresearcher = Agent(agent_name=\"researcher\", model_name=\"gpt-5.4\")\nwriter = Agent(agent_name=\"writer\", model_name=\"gpt-5.4\")\neditor = Agent(agent_name=\"editor\", model_name=\"gpt-5.4\")\n\n# Define a flow: researcher sends work to both writer and editor simultaneously\n# This is a one-to-many relationship\nflow = \"researcher -> writer, editor\"\n\n# Create the rearrangement system\nrearrange_system = AgentRearrange(\n    agents=[researcher, writer, editor],\n    flow=flow,\n)\n\n# Run the swarm\noutputs = rearrange_system.run(\"Analyze the impact of AI on modern cinema.\")\nprint(outputs)\n```\n\n\n\u003C!-- \n### GraphWorkflow\n\n`GraphWorkflow` orchestrates tasks using a Directed Acyclic Graph (DAG), allowing you to manage complex dependencies where some tasks must wait for others to complete.\n\n**Description:** Essential for building sophisticated pipelines, like in software development or complex project management, where task order and dependencies are critical.\n\n```python\nfrom swarms import Agent, GraphWorkflow, Node, Edge, NodeType\n\n# Define agents and a simple python function as nodes\ncode_generator = Agent(agent_name=\"CodeGenerator\", system_prompt=\"Write Python code for the given task.\", model_name=\"gpt-5.4\")\ncode_tester = Agent(agent_name=\"CodeTester\", system_prompt=\"Test the given Python code and find bugs.\", model_name=\"gpt-5.4\")\n\n# Create nodes for the graph\nnode1 = Node(id=\"generator\", agent=code_generator)\nnode2 = Node(id=\"tester\", agent=code_tester)\n\n# Create the graph and define the dependency\ngraph = GraphWorkflow()\ngraph.add_nodes([node1, node2])\ngraph.add_edge(Edge(source=\"generator\", target=\"tester\")) # Tester runs after generator\n\n# Set entry and end points\ngraph.set_entry_points([\"generator\"])\ngraph.set_end_points([\"tester\"])\n\n# Run the graph workflow\nresults = graph.run(\"Create a function that calculates the factorial of a number.\")\nprint(results)\n``` -->\n\n----\n\n### SwarmRouter: The Universal Swarm Orchestrator\n\nThe `SwarmRouter` simplifies building complex workflows by providing a single interface to run any type of swarm. Instead of importing and managing different swarm classes, you can dynamically select the one you need just by changing the `swarm_type` parameter. [Read the full documentation](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fswarm_router\u002F)\n\nThis makes your code cleaner and more flexible, allowing you to switch between different multi-agent strategies with ease. Here's a complete example that shows how to define agents and then use `SwarmRouter` to execute the same task using different collaborative strategies.\n\n```python\nfrom swarms import Agent\nfrom swarms.structs.swarm_router import SwarmRouter, SwarmType\n\n# Define a few generic agents\nwriter = Agent(agent_name=\"Writer\", system_prompt=\"You are a creative writer.\", model_name=\"gpt-5.4\")\neditor = Agent(agent_name=\"Editor\", system_prompt=\"You are an expert editor for stories.\", model_name=\"gpt-5.4\")\nreviewer = Agent(agent_name=\"Reviewer\", system_prompt=\"You are a final reviewer who gives a score.\", model_name=\"gpt-5.4\")\n\n# The agents and task will be the same for all examples\nagents = [writer, editor, reviewer]\ntask = \"Write a short story about a robot who discovers music.\"\n\n# --- Example 1: SequentialWorkflow ---\n# Agents run one after another in a chain: Writer -> Editor -> Reviewer.\nprint(\"Running a Sequential Workflow...\")\nsequential_router = SwarmRouter(swarm_type=SwarmType.SequentialWorkflow, agents=agents)\nsequential_output = sequential_router.run(task)\nprint(f\"Final Sequential Output:\\n{sequential_output}\\n\")\n\n# --- Example 2: ConcurrentWorkflow ---\n# All agents receive the same initial task and run at the same time.\nprint(\"Running a Concurrent Workflow...\")\nconcurrent_router = SwarmRouter(swarm_type=SwarmType.ConcurrentWorkflow, agents=agents)\nconcurrent_outputs = concurrent_router.run(task)\n# This returns a dictionary of each agent's output\nfor agent_name, output in concurrent_outputs.items():\n    print(f\"Output from {agent_name}:\\n{output}\\n\")\n\n# --- Example 3: MixtureOfAgents ---\n# All agents run in parallel, and a special 'aggregator' agent synthesizes their outputs.\nprint(\"Running a Mixture of Agents Workflow...\")\naggregator = Agent(\n    agent_name=\"Aggregator\",\n    system_prompt=\"Combine the story, edits, and review into a final document.\",\n    model_name=\"gpt-5.4\"\n)\nmoa_router = SwarmRouter(\n    swarm_type=SwarmType.MixtureOfAgents,\n    agents=agents,\n    aggregator_agent=aggregator, # MoA requires an aggregator\n)\naggregated_output = moa_router.run(task)\nprint(f\"Final Aggregated Output:\\n{aggregated_output}\\n\")\n```\n\n\nThe `SwarmRouter` is a powerful tool for simplifying multi-agent orchestration. It provides a consistent and flexible way to deploy different collaborative strategies, allowing you to build more sophisticated applications with less code.\n\n-------\n\n### MixtureOfAgents (MoA)\n\nThe `MixtureOfAgents` architecture processes tasks by feeding them to multiple \"expert\" agents in parallel. Their diverse outputs are then synthesized by an aggregator agent to produce a final, high-quality result. [Learn more here](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fmoa_example\u002F)\n\n```python\nfrom swarms import Agent, MixtureOfAgents\n\n# Define expert agents\nfinancial_analyst = Agent(agent_name=\"FinancialAnalyst\", system_prompt=\"Analyze financial data.\", model_name=\"gpt-5.4\")\nmarket_analyst = Agent(agent_name=\"MarketAnalyst\", system_prompt=\"Analyze market trends.\", model_name=\"gpt-5.4\")\nrisk_analyst = Agent(agent_name=\"RiskAnalyst\", system_prompt=\"Analyze investment risks.\", model_name=\"gpt-5.4\")\n\n# Define the aggregator agent\naggregator = Agent(\n    agent_name=\"InvestmentAdvisor\",\n    system_prompt=\"Synthesize the financial, market, and risk analyses to provide a final investment recommendation.\",\n    model_name=\"gpt-5.4\"\n)\n\n# Create the MoA swarm\nmoa_swarm = MixtureOfAgents(\n    agents=[financial_analyst, market_analyst, risk_analyst],\n    aggregator_agent=aggregator,\n)\n\n# Run the swarm\nrecommendation = moa_swarm.run(\"Should we invest in NVIDIA stock right now?\")\nprint(recommendation)\n```\n\n----\n\n### GroupChat\n\n`GroupChat` creates a conversational environment where multiple agents can interact, discuss, and collaboratively solve a problem. You can define the speaking order or let it be determined dynamically. This architecture is ideal for tasks that benefit from debate and multi-perspective reasoning, such as contract negotiation, brainstorming, or complex decision-making.\n\n```python\nfrom swarms import Agent, GroupChat\n\n# Define agents for a debate\ntech_optimist = Agent(agent_name=\"TechOptimist\", system_prompt=\"Argue for the benefits of AI in society.\", model_name=\"gpt-5.4\")\ntech_critic = Agent(agent_name=\"TechCritic\", system_prompt=\"Argue against the unchecked advancement of AI.\", model_name=\"gpt-5.4\")\n\n# Create the group chat\nchat = GroupChat(\n    agents=[tech_optimist, tech_critic],\n    max_loops=4, # Limit the number of turns in the conversation\n)\n\n# Run the chat with an initial topic\nconversation_history = chat.run(\n    \"Let's discuss the societal impact of artificial intelligence.\"\n)\n\n# Print the full conversation\nfor message in conversation_history:\n    print(f\"[{message['agent_name']}]: {message['content']}\")\n```\n\n----\n\n### HierarchicalSwarm\n\n`HierarchicalSwarm` implements a director-worker pattern where a central director agent creates comprehensive plans and distributes specific tasks to specialized worker agents. The director evaluates results and can issue new orders in feedback loops, making it ideal for complex project management and team coordination scenarios.\n\n```python\nfrom swarms import Agent, HierarchicalSwarm\n\n# Define specialized worker agents\ncontent_strategist = Agent(\n    agent_name=\"Content-Strategist\",\n    system_prompt=\"You are a senior content strategist. Develop comprehensive content strategies, editorial calendars, and content roadmaps.\",\n    model_name=\"gpt-5.4\"\n)\n\ncreative_director = Agent(\n    agent_name=\"Creative-Director\", \n    system_prompt=\"You are a creative director. Develop compelling advertising concepts, visual directions, and campaign creativity.\",\n    model_name=\"gpt-5.4\"\n)\n\nseo_specialist = Agent(\n    agent_name=\"SEO-Specialist\",\n    system_prompt=\"You are an SEO expert. Conduct keyword research, optimize content, and develop organic growth strategies.\",\n    model_name=\"gpt-5.4\"\n)\n\nbrand_strategist = Agent(\n    agent_name=\"Brand-Strategist\",\n    system_prompt=\"You are a brand strategist. Develop brand positioning, identity systems, and market differentiation strategies.\",\n    model_name=\"gpt-5.4\"\n)\n\n# Create the hierarchical swarm with a director\nmarketing_swarm = HierarchicalSwarm(\n    name=\"Marketing-Team-Swarm\",\n    description=\"A comprehensive marketing team with specialized agents coordinated by a director\",\n    agents=[content_strategist, creative_director, seo_specialist, brand_strategist],\n    max_loops=2,  # Allow for feedback and refinement\n    verbose=True\n)\n\n# Run the swarm on a complex marketing challenge\nresult = marketing_swarm.run(\n    \"Develop a comprehensive marketing strategy for a new SaaS product launch. \"\n    \"The product is a project management tool targeting small to medium businesses. \"\n    \"Coordinate the team to create content strategy, creative campaigns, SEO optimization, \"\n    \"and brand positioning that work together cohesively.\"\n)\n\nprint(result)\n```\n\nThe `HierarchicalSwarm` excels at:\n- **Complex Project Management**: Breaking down large tasks into specialized subtasks\n- **Team Coordination**: Ensuring all agents work toward unified goals\n- **Quality Control**: Director provides feedback and refinement loops\n- **Scalable Workflows**: Easy to add new specialized agents as needed\n\n---\n\n### HeavySwarm\n\n`HeavySwarm` implements a sophisticated 5-phase workflow inspired by X.AI's Grok heavy implementation. It uses specialized agents (Research, Analysis, Alternatives, Verification) to provide comprehensive task analysis through intelligent question generation, parallel execution, and synthesis. This architecture excels at complex research and analysis tasks requiring thorough investigation and multiple perspectives.\n\n```python\nfrom swarms import HeavySwarm\n\n# Pip install swarms-tools\nfrom swarms_tools import exa_search\n\nswarm = HeavySwarm(\n    name=\"Gold ETF Research Team\",\n    description=\"A team of agents that research the best gold ETFs\",\n    worker_model_name=\"claude-sonnet-4-20250514\",\n    show_dashboard=True,\n    question_agent_model_name=\"gpt-4.1\",\n    loops_per_agent=1,\n    agent_prints_on=False,\n    worker_tools=[exa_search],\n    random_loops_per_agent=True,\n)\n\nprompt = (\n    \"Find the best 3 gold ETFs. For each ETF, provide the ticker symbol, \"\n    \"full name, current price, expense ratio, assets under management, and \"\n    \"a brief explanation of why it is considered among the best. Present the information \"\n    \"in a clear, structured format suitable for investors. Scrape the data from the web. \"\n)\n\nout = swarm.run(prompt)\nprint(out)\n\n```\n\nThe `HeavySwarm` provides:\n\n- **5-Phase Analysis**: Question generation, research, analysis, alternatives, and verification\n\n- **Specialized Agents**: Each phase uses purpose-built agents for optimal results\n\n- **Comprehensive Coverage**: Multiple perspectives and thorough investigation\n\n- **Real-time Dashboard**: Optional visualization of the analysis process\n\n- **Structured Output**: Well-organized and actionable results\n\nThis architecture is perfect for financial analysis, strategic planning, research reports, and any task requiring deep, multi-faceted analysis. [Learn more about HeavySwarm](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fheavy_swarm\u002F)\n\n---\n\n### MAKER\n\n`MAKER` implements **maximal agentic decomposition** with **first-to-ahead-by-k voting** and **red-flagging**: you supply `format_prompt`, `parse_response`, and optional `validate_response` \u002F `update_state`, then run for a fixed number of steps (or until a stop condition). Each step spins up a focused one-shot `Agent` (or cycles a pool you provide) until one parsed answer leads all others by `k` votes. This matches the error-correction story in [Solving a Million-Step LLM Task with Zero Errors](https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.09030). [Full documentation](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fmaker\u002F)\n\n```python\nfrom swarms.structs.maker import MAKER\n\nmaker = MAKER(\n    model_name=\"gpt-4.1-mini\",\n    system_prompt=\"You solve tasks in one clear line per step.\",\n    k=3,\n)\n\n# Optional: override format_prompt \u002F parse_response \u002F validate_response for your domain.\nresults = maker.run(\n    task=\"List three concise benefits of typed APIs, one per step.\",\n    max_steps=3,\n)\nprint(results)\n```\n\nFor lower latency when `k` is large, use `run_parallel_voting` with the same `task` and `max_steps`.\n\n---\n\n### Social Algorithms\n\n**Social Algorithms** provide a flexible framework for defining custom communication patterns between agents. You can upload any arbitrary social algorithm as a callable that defines the sequence of communication, enabling agents to talk to each other in sophisticated ways. [Learn more about Social Algorithms](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fsocial_algorithms\u002F)\n\n```python\nfrom swarms import Agent, SocialAlgorithms\n\n# Define a custom social algorithm\ndef research_analysis_synthesis_algorithm(agents, task, **kwargs):\n    # Agent 1 researches the topic\n    research_result = agents[0].run(f\"Research: {task}\")\n    \n    # Agent 2 analyzes the research\n    analysis = agents[1].run(f\"Analyze this research: {research_result}\")\n    \n    # Agent 3 synthesizes the findings\n    synthesis = agents[2].run(f\"Synthesize: {research_result} + {analysis}\")\n    \n    return {\n        \"research\": research_result,\n        \"analysis\": analysis,\n        \"synthesis\": synthesis\n    }\n\n# Create agents\nresearcher = Agent(\n  agent_name=\"Researcher\",\n  agent_description=\"Expert in comprehensive research and information gathering.\",\n  model_name=\"gpt-4.1\"\n)\nanalyst = Agent(\n  agent_name=\"Analyst\",\n  agent_description=\"Specialist in analyzing and interpreting data.\",\n  model_name=\"gpt-4.1\"\n)\nsynthesizer = Agent(\n  agent_name=\"Synthesizer\",\n  agent_description=\"Focused on synthesizing and integrating research insights.\",\n  model_name=\"gpt-4.1\"\n)\n\n# Create social algorithm\nsocial_alg = SocialAlgorithms(\n    name=\"Research-Analysis-Synthesis\",\n    agents=[researcher, analyst, synthesizer],\n    social_algorithm=research_analysis_synthesis_algorithm,\n    verbose=True\n)\n\n# Run the algorithm\nresult = social_alg.run(\"The impact of AI on healthcare\")\nprint(result.final_outputs)\n```\n\nPerfect for implementing complex multi-agent workflows, collaborative problem-solving, and custom communication protocols.\n\n---\n\n### Agent Orchestration Protocol (AOP)\n\nThe **Agent Orchestration Protocol (AOP)** is a powerful framework for deploying and managing agents as distributed services. AOP enables agents to be discovered, managed, and executed through a standardized protocol, making it perfect for building scalable multi-agent systems. [Learn more about AOP](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Faop\u002F)\n\n```python\nfrom swarms import Agent\nfrom swarms.structs.aop import AOP\n\n# Create specialized agents\nresearch_agent = Agent(\n    agent_name=\"Research-Agent\",\n    agent_description=\"Expert in research and data collection\",\n    model_name=\"anthropic\u002Fclaude-sonnet-4-5\",\n    max_loops=1,\n    tags=[\"research\", \"data-collection\", \"analysis\"],\n    capabilities=[\"web-search\", \"data-gathering\", \"report-generation\"],\n    role=\"researcher\"\n)\n\nanalysis_agent = Agent(\n    agent_name=\"Analysis-Agent\", \n    agent_description=\"Expert in data analysis and insights\",\n    model_name=\"anthropic\u002Fclaude-sonnet-4-5\",\n    max_loops=1,\n    tags=[\"analysis\", \"data-processing\", \"insights\"],\n    capabilities=[\"statistical-analysis\", \"pattern-recognition\", \"visualization\"],\n    role=\"analyst\"\n)\n\n# Create AOP server\ndeployer = AOP(\n    server_name=\"ResearchCluster\",\n    port=8000,\n    verbose=True\n)\n\n# Add agents to the server\ndeployer.add_agent(\n    agent=research_agent,\n    tool_name=\"research_tool\",\n    tool_description=\"Research and data collection tool\",\n    timeout=30,\n    max_retries=3\n)\n\ndeployer.add_agent(\n    agent=analysis_agent,\n    tool_name=\"analysis_tool\", \n    tool_description=\"Data analysis and insights tool\",\n    timeout=30,\n    max_retries=3\n)\n\n# List all registered agents\nprint(\"Registered agents:\", deployer.list_agents())\n\n# Start the AOP server\ndeployer.run()\n```\n\nPerfect for deploying large scale multi-agent systems. [Read the complete AOP documentation](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Faop\u002F)\n\n---\n\n## Documentation\n\nDocumentation is located here at: [docs.swarms.world](https:\u002F\u002Fdocs.swarms.world)\n\n\n---\n\n## Examples\n\nExplore comprehensive examples and tutorials to learn how to use Swarms effectively.\n\n| Category | Example | Description | Link |\n|----------|---------|-------------|------|\n| **Basic Examples** | Basic Agent | Simple agent setup and usage | [Basic Agent](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fbasic_agent\u002F) |\n| **Basic Examples** | Agent with Tools | Using agents with various tools | [Agent with Tools](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fagent_with_tools\u002F) |\n| **Basic Examples** | Agent with Structured Outputs | Working with structured data outputs | [Structured Outputs](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fagent_structured_outputs\u002F) |\n| **Basic Examples** | Agent with MCP Integration | Model Context Protocol integration | [MCP Integration](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fagent_with_mcp\u002F) |\n| **Basic Examples** | Vision Processing | Agents with image processing capabilities | [Vision Processing](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fvision_processing\u002F) |\n| **Basic Examples** | Multiple Images | Working with multiple images | [Multiple Images](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fmultiple_images\u002F) |\n| **Basic Examples** | Vision and Tools | Combining vision with tool usage | [Vision and Tools](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fvision_tools\u002F) |\n| **Basic Examples** | Agent Streaming | Real-time agent output streaming | [Agent Streaming](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fexamples\u002Fagent_stream\u002F) |\n| **Basic Examples** | Agent Output Types | Different output formats and types | [Output Types](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fagent_output_types\u002F) |\n| **Basic Examples** | Gradio Chat Interface | Building interactive chat interfaces | [Gradio UI](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fui\u002Fmain\u002F) |\n| **Model Providers** | Model Providers Overview | Complete guide to supported models | [Model Providers](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fmodel_providers\u002F) |\n| **Model Providers** | OpenAI | OpenAI model integration | [OpenAI Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fopenai_example\u002F) |\n| **Model Providers** | Anthropic | Claude model integration | [Anthropic Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fclaude\u002F) |\n| **Model Providers** | Groq | Groq model integration | [Groq Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fgroq\u002F) |\n| **Model Providers** | Cohere | Cohere model integration | [Cohere Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fcohere\u002F) |\n| **Model Providers** | DeepSeek | DeepSeek model integration | [DeepSeek Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fdeepseek\u002F) |\n| **Model Providers** | Ollama | Local Ollama model integration | [Ollama Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Follama\u002F) |\n| **Model Providers** | OpenRouter | OpenRouter model integration | [OpenRouter Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fopenrouter\u002F) |\n| **Model Providers** | XAI | XAI model integration | [XAI Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fxai\u002F) |\n| **Model Providers** | Llama4 | Llama4 model integration | [Llama4 Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fllama4\u002F) |\n| **Multi-Agent Architecture** | HierarchicalSwarm | Hierarchical agent orchestration | [HierarchicalSwarm Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fhierarchical_swarm_example\u002F) |\n| **Multi-Agent Architecture** | Hybrid Hierarchical-Cluster Swarm | Advanced hierarchical patterns | [HHCS Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fhhcs_examples\u002F) |\n| **Multi-Agent Architecture** | GroupChat | Multi-agent conversations | [GroupChat Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fgroupchat_example\u002F) |\n| **Multi-Agent Architecture** | Sequential Workflow | Step-by-step agent workflows | [Sequential Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fsequential_example\u002F) |\n| **Multi-Agent Architecture** | SwarmRouter | Universal swarm orchestration | [SwarmRouter Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fswarm_router\u002F) |\n| **Multi-Agent Architecture** | MultiAgentRouter | Minimal router example | [MultiAgentRouter Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fmulti_agent_router_minimal\u002F) |\n| **Multi-Agent Architecture** | ConcurrentWorkflow | Parallel agent execution | [Concurrent Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fconcurrent_workflow\u002F) |\n| **Multi-Agent Architecture** | Mixture of Agents | Expert agent collaboration | [MoA Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fmoa_example\u002F) |\n| **Multi-Agent Architecture** | Unique Swarms | Specialized swarm patterns | [Unique Swarms](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Funique_swarms\u002F) |\n| **Multi-Agent Architecture** | Agents as Tools | Using agents as tools in workflows | [Agents as Tools](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fagents_as_tools\u002F) |\n| **Multi-Agent Architecture** | Aggregate Responses | Combining multiple agent outputs | [Aggregate Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Faggregate\u002F) |\n| **Multi-Agent Architecture** | Interactive GroupChat | Real-time agent interactions | [Interactive GroupChat](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Figc_example\u002F) |\n| **Deployment Solutions** | Agent Orchestration Protocol (AOP) | Deploy agents as distributed services with discovery and management | [AOP Reference](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Faop\u002F) |\n| **Applications** | Advanced Research System | Multi-agent research system inspired by Anthropic's research methodology | [AdvancedResearch](https:\u002F\u002Fgithub.com\u002FThe-Swarm-Corporation\u002FAdvancedResearch) |\n| **Applications** | Hospital Simulation | Healthcare simulation system using multi-agent architecture | [HospitalSim](https:\u002F\u002Fgithub.com\u002FThe-Swarm-Corporation\u002FHospitalSim) |\n| **Applications** | Browser Agents | Web automation with agents | [Browser Agents](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fswarms_of_browser_agents\u002F) |\n| **Applications** | Medical Analysis | Healthcare applications | [Medical Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fswarms_api_medical\u002F) |\n| **Applications** | Finance Analysis | Financial applications | [Finance Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fswarms_api_finance\u002F) |\n| **Cookbook & Templates** | Examples Overview | Complete examples directory | [Examples Index](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fexamples\u002F) |\n| **Cookbook & Templates** | Cookbook Index | Curated example collection | [Cookbook](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fexamples\u002Fcookbook_index\u002F) |\n| **Cookbook & Templates** | Paper Implementations | Research paper implementations | [Paper Implementations](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fexamples\u002Fpaper_implementations\u002F) |\n| **Cookbook & Templates** | Templates & Applications | Reusable templates | [Templates](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fexamples\u002Ftemplates\u002F) |\n\n---\n\n## Contribute to Swarms\n\nOur mission is to accelerate the transition to a fully autonomous world economy by providing enterprise-grade, production-ready infrastructure that enables seamless deployment and orchestration of millions of autonomous agents. We are creating the operating system for the agent economy, and we need your help to achieve this goal.\n\nSwarms is built by the community, for the community. We believe that collaborative development is the key to pushing the boundaries of what's possible with multi-agent AI. Your contributions are not only welcome—they are essential to our mission. [Learn more about why you should contribute to Swarms](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fcontributors\u002Fmain\u002F)\n\n### Why Contribute?\n\nBy joining us, you have the opportunity to:\n\n* **Work on the Frontier of Agents:** Shape the future of autonomous agent technology and help build a production-grade, open-source framework.\n\n* **Join a Vibrant Community:** Collaborate with a passionate and growing group of agent developers, researchers, and agent enthusasits.\n\n* **Make a Tangible Impact:** Whether you're fixing a bug, adding a new feature, or improving documentation, your work will be used in real-world applications.\n\n* **Learn and Grow:** Gain hands-on experience with advanced AI concepts and strengthen your software engineering skills.\n\nDiscover more about our mission and the benefits of becoming a contributor in our official [**Contributor's Guide**](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fcontributors\u002Fmain\u002F).\n\n### How to Get Started\n\nWe've made it easy to start contributing. Here's how you can help:\n\n1. **Find an Issue to Tackle:** The best way to begin is by visiting our [**contributing project board**](https:\u002F\u002Fgithub.com\u002Fusers\u002Fkyegomez\u002Fprojects\u002F1). Look for issues tagged with `good first issue`—these are specifically selected for new contributors.\n\n2. **Report a Bug or Request a Feature:** Have a new idea or found something that isn't working right? We'd love to hear from you. Please [**file a Bug Report or Feature Request**](https:\u002F\u002Fgithub.com\u002Fkyegomez\u002Fswarms\u002Fissues) on our GitHub Issues page.\n\n3. **Understand Our Workflow and Standards:** Before submitting your work, please review our complete [**Contribution Guidelines**](https:\u002F\u002Fgithub.com\u002Fkyegomez\u002Fswarms\u002Fblob\u002Fmaster\u002FCONTRIBUTING.md). To help maintain code quality, we also encourage you to read our guide on [**Code Cleanliness**](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fframework\u002Fcode_cleanliness\u002F).\n\n4. **Join the Discussion:** To participate in roadmap discussions and connect with other developers, join our community on [**Discord**](https:\u002F\u002Fdiscord.gg\u002FEamjgSaEQf).\n\n\n### Thank You to Our Contributors\n\nThank you for contributing to swarms. Your work is extremely appreciated and recognized.\n\n\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fkyegomez\u002Fswarms\u002Fgraphs\u002Fcontributors\">\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fkyegomez_swarms_readme_4afdaf65db2f.png\" \u002F>\n\u003C\u002Fa>\n\n### Thank You to Our Community\n\nWe're incredibly grateful to everyone who supports Swarms! Your stars, forks, and contributions help make this project better every day.\n\n[![Forkers repo roster for @kyegomez\u002Fswarms](https:\u002F\u002Freporoster.com\u002Fforks\u002Fkyegomez\u002Fswarms)](https:\u002F\u002Fgithub.com\u002Fkyegomez\u002Fswarms\u002Fnetwork\u002Fmembers)\n\n[![Stargazers repo roster for @kyegomez\u002Fswarms](https:\u002F\u002Freporoster.com\u002Fstars\u002Fkyegomez\u002Fswarms)](https:\u002F\u002Fgithub.com\u002Fkyegomez\u002Fswarms\u002Fstargazers)\n\n-----\n\n## Join the Swarms community\n\nJoin our community of agent engineers and researchers for technical support, cutting-edge updates, and exclusive access to world-class agent engineering insights!\n\n| Platform | Description | Link |\n|----------|-------------|------|\n| Documentation | Official documentation and guides | [docs.swarms.world](https:\u002F\u002Fdocs.swarms.world) |\n| Blog | Latest updates and technical articles | [Medium](https:\u002F\u002Fmedium.com\u002F@kyeg) |\n| Discord | Live chat and community support | [Join Discord](https:\u002F\u002Fdiscord.gg\u002FEamjgSaEQf) |\n| Twitter | Latest news and announcements | [@swarms_corp](https:\u002F\u002Ftwitter.com\u002Fswarms_corp) |\n| LinkedIn | Professional network and updates | [The Swarm Corporation](https:\u002F\u002Fwww.linkedin.com\u002Fcompany\u002Fthe-swarm-corporation) |\n| YouTube | Tutorials and demos | [Swarms Channel](https:\u002F\u002Fwww.youtube.com\u002Fchannel\u002FUC9yXyitkbU_WSy7bd_41SqQ) |\n| Events | Join our community events | [Sign up here](https:\u002F\u002Flu.ma\u002Fswarms_calendar) |\n| Onboarding Session | Get onboarded with Kye Gomez, creator and lead maintainer of Swarms | [Book Session](https:\u002F\u002Fcal.com\u002Fswarms\u002Fswarms-onboarding-session) |\n\n------\n\n## Citation\n\nIf you use **swarms** in your research, please cite the project by referencing the metadata in [CITATION.cff](.\u002FCITATION.cff).\n\n```bibtex\n@misc{SWARMS_2022,\n  author  = {Kye Gomez and Pliny and Zack Bradshaw and Ilumn and Harshal and the Swarms Community},\n  title   = {{Swarms: Production-Grade Multi-Agent Infrastructure Platform}},\n  year    = {2022},\n  howpublished = {\\url{https:\u002F\u002Fgithub.com\u002Fkyegomez\u002Fswarms}},\n  note    = {Documentation available at \\url{https:\u002F\u002Fdocs.swarms.world}},\n  version = {latest}\n```\n\n---\n\n# License\n\nSwarms is licensed under the Apache License 2.0. [Learn more here](.\u002FLICENSE)\n","\u003Cdiv align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fswarms.world\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fkyegomez_swarms_readme_1df629ea353f.png\" style=\"margin: 15px; max-width: 350px\" width=\"80%\" alt=\"Logo\">\n  \u003C\u002Fa>\n\u003C\u002Fdiv>\n\u003Cp align=\"center\">\n  \u003Cem>企业级、生产就绪的多智能体（Multi-Agent）编排框架\u003C\u002Fem>\n\u003C\u002Fp>\n\n\n\u003Cp align=\"center\">\n  \u003C!-- Main Navigation Links -->\n  \u003Ca href=\"https:\u002F\u002Fswarms.ai\">Swarms 官网\u003C\u002Fa>\n  \u003Cspan>  •  \u003C\u002Fspan>\n  \u003Ca href=\"https:\u002F\u002Fdocs.swarms.world\">文档\u003C\u002Fa>\n  \u003Cspan>  •  \u003C\u002Fspan>\n  \u003Ca href=\"https:\u002F\u002Fswarms.world\">Swarms 市场\u003C\u002Fa>\n\u003C\u002Fp>\n\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fpypi.org\u002Fproject\u002Fswarms\u002F\" target=\"_blank\">\n    \u003Cpicture>\n      \u003Csource srcset=\"https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fv\u002Fswarms?style=for-the-badge&color=3670A0\" media=\"(prefers-color-scheme: dark)\">\n      \u003Cimg alt=\"Version\" src=\"https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fv\u002Fswarms?style=for-the-badge&color=3670A0\">\n    \u003C\u002Fpicture>\n  \u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fpypi.org\u002Fproject\u002Fswarms\u002F\" target=\"_blank\">\n    \u003Cpicture>\n      \u003Csource srcset=\"https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fdm\u002Fswarms?style=for-the-badge&color=3670A0\" media=\"(prefers-color-scheme: dark)\">\n      \u003Cimg alt=\"Downloads\" src=\"https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fdm\u002Fswarms?style=for-the-badge&color=3670A0\">\n    \u003C\u002Fpicture>\n  \u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Ftwitter.com\u002Fswarms_corp\u002F\">\n    \u003Cpicture>\n      \u003Csource srcset=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FTwitter-Follow-1DA1F2?style=for-the-badge&logo=twitter&logoColor=white\" media=\"(prefers-color-scheme: dark)\">\n      \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FTwitter-Follow-1DA1F2?style=for-the-badge&logo=twitter&logoColor=white\" alt=\"Twitter\">\n    \u003C\u002Fpicture>\n  \u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fdiscord.gg\u002FEamjgSaEQf\">\n    \u003Cpicture>\n      \u003Csource srcset=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDiscord-Join-5865F2?style=for-the-badge&logo=discord&logoColor=white\" media=\"(prefers-color-scheme: dark)\">\n      \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDiscord-Join-5865F2?style=for-the-badge&logo=discord&logoColor=white\" alt=\"Discord\">\n    \u003C\u002Fpicture>\n  \u003C\u002Fa>\n\u003C\u002Fp>\n\n## 功能特性\n\nSwarms 提供了一个全面的企业级多智能体（Multi-Agent）基础设施平台，专为生产环境的大规模部署和与现有系统的无缝集成而设计。[在此了解更多关于 Swarms 的功能特性](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Ffeatures\u002F)\n\n| 类别 | 功能特性 | 优势 |\n|----------|----------|-----------|\n| **企业级架构** | • 生产就绪的基础设施\u003Cbr>• 高可用性系统\u003Cbr>• 模块化微服务设计\u003Cbr>• 全面的可观测性（Observability）\u003Cbr>• 向后兼容性 | • 99.9%+ 的正常运行时间保障\u003Cbr>• 降低运维开销\u003Cbr>• 无缝集成遗留系统\u003Cbr>• 增强的系统监控能力\u003Cbr>• 无风险迁移路径 |\n| **多智能体编排** | • 分层智能体集群（Hierarchical Agent Swarms）\u003Cbr>• 并行处理流水线\u003Cbr>• 顺序工作流编排\u003Cbr>• 基于图的智能体网络\u003Cbr>• 动态智能体组合\u003Cbr>• 智能体注册中心管理 | • 复杂业务流程自动化\u003Cbr>• 可扩展的任务分发\u003Cbr>• 灵活的工作流适配\u003Cbr>• 优化的资源利用率\u003Cbr>• 集中式智能体治理\u003Cbr>• 企业级智能体生命周期管理 |\n| **企业集成能力** | • 多模型提供商支持\u003Cbr>• 自定义智能体开发框架\u003Cbr>• 丰富的企用工具库\u003Cbr>• 多种记忆系统（Memory Systems）\u003Cbr>• 向后兼容 LangChain、AutoGen、CrewAI\u003Cbr>• 标准化 API 接口 | • 与供应商无关的架构\u003Cbr>• 支持定制化解决方案开发\u003Cbr>• 扩展功能集成\u003Cbr>• 增强的知识管理\u003Cbr>• 无缝框架迁移\u003Cbr>• 降低集成复杂度 |\n| **企业级可扩展性** | • 并发多智能体处理\u003Cbr>• 智能资源管理\u003Cbr>• 负载均衡与自动扩缩容\u003Cbr>• 水平扩展能力\u003Cbr>• 性能优化\u003Cbr>• 容量规划工具 | • 高吞吐量处理\u003Cbr>• 成本效益高的资源利用\u003Cbr>• 按需弹性伸缩\u003Cbr>• 线性性能扩展\u003Cbr>• 优化的响应时间\u003Cbr>• 可预测的增长规划 |\n| **开发者体验** | • 直观的企业级 API\u003Cbr>• 全面的文档\u003Cbr>• 活跃的企业社区\u003Cbr>• CLI 与 SDK 工具\u003Cbr>• IDE 集成支持\u003Cbr>• 代码生成模板 | • 加速开发周期\u003Cbr>• 降低学习曲线\u003Cbr>• 专家社区支持\u003Cbr>• 快速部署能力\u003Cbr>• 提升开发者生产力\u003Cbr>• 标准化的开发模式 |\n\n## 支持的协议与集成\n\nSwarms 无缝集成了行业标准协议和开放规范，为工具集成、支付处理、分布式智能体编排（Agent Orchestration）以及模型互操作性（Model Interoperability）解锁了强大能力。\n\n| 协议 | 描述 | 使用场景 | 文档 |\n|----------|-------------|-----------|---------------|\n| **[MCP (Model Context Protocol)](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fmulti_mcp_agent\u002F)** | 一种标准化协议，允许 AI 智能体通过 MCP 服务器与外部工具和服务交互，支持动态工具发现与执行。 | • 工具集成\u003Cbr>• 多服务器连接\u003Cbr>• 外部 API 访问\u003Cbr>• 数据库连接 | [MCP 集成指南](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fmulti_mcp_agent\u002F) |\n| **[X402](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fexamples\u002Fx402_payment_integration\u002F)** | 一种面向 API 端点的加密货币支付协议，支持按使用付费（pay-per-use）模式对智能体进行变现。 | • 智能体变现\u003Cbr>• 支付网关保护\u003Cbr>• 加密货币支付\u003Cbr>• 按使用付费服务 | [X402 快速入门](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fexamples\u002Fx402_payment_integration\u002F) |\n| **[AOP (Agent Orchestration Protocol)](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fexamples\u002Faop_medical\u002F)** | 用于将智能体作为分布式服务部署和管理的框架，通过标准化协议实现智能体发现、管理和执行。 | • 分布式智能体部署\u003Cbr>• 智能体发现\u003Cbr>• 服务编排\u003Cbr>• 可扩展的多智能体系统 | [AOP 参考文档](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Faop\u002F) |\n| **[Swarms Marketplace](https:\u002F\u002Fswarms.world)** | 一个用于发现和共享生产就绪型提示（prompts）、智能体和工具的平台，支持从市场自动加载提示，并可直接从代码中发布自己的提示。 | • 提示发现与复用\u003Cbr>• 一行代码加载提示\u003Cbr>• 社区提示共享\u003Cbr>• 提示变现 | [Marketplace 教程](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fmarketplace_prompt_loading\u002F) |\n| **[Open Responses](https:\u002F\u002Fwww.openresponses.org\u002F)** | 基于 OpenAI Responses API 的开源规范和生态系统，提供跨提供商、可互操作的大语言模型（LLM）接口。它定义了统一的数据结构和工具，用于调用语言模型、流式输出结果以及构建智能体工作流，且不依赖于特定提供商。 | • 统一的 LLM 接口\u003Cbr>• 流式输出\u003Cbr>• 多提供商编排\u003Cbr>• 可互操作的智能体工作流 | [Open Responses 官网](https:\u002F\u002Fwww.openresponses.org\u002F) |\n| **[Agent Skills](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fagents\u002Fagent_skills\u002F)** | 由 Anthropic 提出的一种轻量级、基于 Markdown 的格式，用于定义模块化、可复用的智能体能力。通过加载简单的 SKILL.md 文件即可实现智能体专业化，无需修改代码。 | • 智能体专业化\u003Cbr>• 可复用技能库\u003Cbr>• 无代码智能体定制\u003Cbr>• 兼容 Claude Code | [Agent Skills 文档](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fagents\u002Fagent_skills\u002F) |\n\n\n## 安装\n\n### 使用 pip\n\n```bash\n$ pip3 install -U swarms\n```\n\n### 使用 uv（推荐）\n\n[uv](https:\u002F\u002Fgithub.com\u002Fastral-sh\u002Fuv) 是一个用 Rust 编写的快速 Python 包安装器和依赖解析器。\n\n```bash\n$ uv pip install swarms\n```\n\n### 使用 poetry\n\n```bash\n$ poetry add swarms\n```\n\n### 从源码安装\n\n```bash\n# 克隆仓库\n$ git clone https:\u002F\u002Fgithub.com\u002Fkyegomez\u002Fswarms.git\n$ cd swarms\n$ pip install -r requirements.txt\n```\n\n\u003C!-- ### 使用 Docker\n\n使用我们预构建的 Docker 镜像是开始使用 Swarms 的最简单方式：\n\n```bash\n# 拉取并运行最新镜像\n$ docker pull kyegomez\u002Fswarms:latest\n$ docker run --rm kyegomez\u002Fswarms:latest python -c \"import swarms; print('Swarms is ready!')\"\n\n# 以交互模式运行（适用于开发）\n$ docker run -it --rm -v $(pwd):\u002Fapp kyegomez\u002Fswarms:latest bash\n\n# 使用 docker-compose（推荐用于开发）\n$ docker-compose up -d\n```\n\n有关更多 Docker 选项和高级用法，请参阅我们的 [Docker 文档](\u002Fscripts\u002Fdocker\u002FDOCKER.md)。 -->\n\n---\n\n## 环境配置\n\n[在此了解更多关于环境配置的信息](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Finstall\u002Fenv\u002F)\n\n```\nOPENAI_API_KEY=\"\"\nWORKSPACE_DIR=\"agent_workspace\"\nANTHROPIC_API_KEY=\"\"\nGROQ_API_KEY=\"\"\n```\n\n\n### 你的第一个智能体（Agent）\n\n**智能体（Agent）** 是 swarm 的基本构建单元——一个由大语言模型（LLM）+ 工具（Tools）+ 记忆（Memory）驱动的自主实体。[了解更多](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fagent\u002F)\n\n```python\nfrom swarms import Agent\n\n# 初始化一个新智能体\nagent = Agent(\n    model_name=\"gpt-5.4\", # 指定 LLM\n    max_loops=\"auto\",              # 设置交互次数\n    interactive=True,         # 启用交互模式以获得实时反馈\n)\n\n# 让智能体执行任务\nagent.run(\"使用多智能体系统的主要优势有哪些？\")\n```\n\n### 你的第一个 Swarm：多智能体协作\n\n**Swarm** 由多个协同工作的智能体组成。以下简单示例创建了一个包含两个智能体的工作流，用于研究并撰写一篇博客文章。[了解更多关于 SequentialWorkflow](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fsequential_workflow\u002F)\n\n```python\nfrom swarms import Agent, SequentialWorkflow\n\n# 智能体 1：研究员\nresearcher = Agent(\n    agent_name=\"Researcher\",\n    system_prompt=\"你的任务是研究提供的主题并提供详细摘要。\",\n    model_name=\"gpt-5.4\",\n)\n\n# 智能体 2：撰稿人\nwriter = Agent(\n    agent_name=\"Writer\",\n    system_prompt=\"你的任务是根据研究摘要撰写一篇优美、引人入胜的博客文章。\",\n    model_name=\"gpt-5.4\",\n)\n\n# 创建一个顺序工作流，研究员的输出将作为撰稿人的输入\nworkflow = SequentialWorkflow(agents=[researcher, writer])\n\n# 在指定任务上运行工作流\nfinal_post = workflow.run(\"人工智能的历史与未来\")\nprint(final_post)\n\n```\n\n-----\n\n### AutoSwarmBuilder：自主智能体生成\n\n`AutoSwarmBuilder` 能根据你的任务描述自动生成专用智能体及其工作流。只需描述你的需求，它就会创建一个完整的多智能体系统，包含详细的提示词和最优的智能体配置。[了解更多关于 AutoSwarmBuilder](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fauto_swarm_builder\u002F)\n\n```python\nfrom swarms.structs.auto_swarm_builder import AutoSwarmBuilder\nimport json\n\n# 初始化 AutoSwarmBuilder\nswarm = AutoSwarmBuilder(\n    name=\"My Swarm\",\n    description=\"A swarm of agents\",\n    verbose=True,\n    max_loops=1,\n    return_agents=True,\n    model_name=\"gpt-5.4\",\n)\n```\n\n# 让构建器自动创建智能体（agents）和工作流（workflows）\n\n```python\nresult = swarm.run(\n    task=\"创建一个会计团队来分析加密货币交易，\"\n         \"团队中必须包含 5 个智能体，且每个智能体的提示词（prompts）必须极其详尽。\"\n         \"确保提示词极度详细、具体、冗长且全面。\"\n         \"务必在提示词中包含任务的所有细节。\"\n)\n\n# result 包含生成的智能体及其配置\nprint(json.dumps(result, indent=4))\n```\n\n`AutoSwarmBuilder` 提供以下功能：\n\n- **自动智能体生成（Automatic Agent Generation）**：根据任务需求创建专业化的智能体  \n- **智能提示工程（Intelligent Prompt Engineering）**：为每个智能体生成全面、详细的提示词  \n- **最优工作流设计（Optimal Workflow Design）**：确定最佳的智能体交互方式和工作流结构  \n- **生产就绪配置（Production-Ready Configurations）**：返回可直接部署的完整配置智能体  \n- **灵活架构（Flexible Architecture）**：支持多种 swarm 类型和智能体专业化方向  \n\n该功能非常适合快速原型开发、复杂任务分解，以及在无需手动配置的情况下创建专业化的智能体团队。\n\n-----\n\n## 可用的多智能体架构（Multi-Agent Architectures）\n\n`swarms` 提供了多种强大且预构建的多智能体架构，使您能够以不同方式编排智能体。为您的特定问题选择合适的结构，以构建高效可靠的生产系统。\n\n| **架构（Architecture）** | **描述（Description）** | **适用场景（Best For）** |\n|---|---|---|\n| **[SequentialWorkflow](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fsequential_workflow\u002F)** | 智能体按线性链式顺序执行任务；前一个智能体的输出作为下一个智能体的输入。 | 分步处理流程，例如数据转换管道和报告生成。 |\n| **[ConcurrentWorkflow](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fconcurrentworkflow\u002F)** | 智能体同时运行任务，以实现最高效率。 | 高吞吐量任务，例如批处理和并行数据分析。 |\n| **[AgentRearrange](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fagent_rearrange\u002F)** | 动态映射智能体之间的复杂关系（例如 `a -> b, c`）。 | 灵活且自适应的工作流、任务分发和动态路由。 |\n| **[GraphWorkflow](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fgraph_workflow\u002F)** | 将智能体作为有向无环图（DAG）中的节点进行编排。 | 具有复杂依赖关系的项目，例如软件构建。 |\n| **[MixtureOfAgents (MoA)](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fmoa\u002F)** | 并行使用多个专家智能体，并综合其输出结果。 | 通过协作实现复杂问题求解和达到前沿性能。 |\n| **[GroupChat](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fgroup_chat\u002F)** | 智能体通过对话界面协作并做出决策。 | 实时协作决策、谈判和头脑风暴。 |\n| **[ForestSwarm](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fforest_swarm\u002F)** | 动态选择最适合当前任务的单个智能体或智能体树。 | 任务路由、基于专业能力的优化以及复杂决策树。 |\n| **[HierarchicalSwarm](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fhierarchical_swarm\u002F)** | 由一个指挥者（director）智能体制定计划，并将任务分发给专业化的工作者（worker）智能体。 | 复杂项目管理、团队协调以及带反馈循环的层级化决策。 |\n| **[HeavySwarm](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fheavy_swarm\u002F)** | 实现五阶段工作流，包含专业化智能体（研究、分析、备选方案、验证），用于全面任务分析。 | 复杂研究与分析任务、财务分析、战略规划和综合报告。 |\n| **[MAKER](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fmaker\u002F)** | 将长期任务分解为多个步骤；每一步使用“first-to-ahead-by-k”投票机制，并对微智能体样本进行红标（red-flagging）检测（源自 Meyerson 等人，2025）。 | 极长或脆弱的流水线任务，要求在每个原子步骤上获得统计一致性与验证——而非人工设计的多智能体图。 |\n| **[SwarmRouter](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fswarm_router\u002F)** | 通用编排器，提供统一接口以动态选择并运行任意类型的 swarm。 | 简化复杂工作流、在不同 swarm 策略间切换以及统一的多智能体管理。 |\n\n-----\n\n### SequentialWorkflow\n\n`SequentialWorkflow` 以严格顺序执行任务，形成一个流水线，其中每个智能体都在前一个智能体工作的基础上继续处理。`SequentialWorkflow` 非常适合具有清晰、有序步骤的流程，确保依赖任务被正确处理。\n\n```python\nfrom swarms import Agent, SequentialWorkflow\n\n# 智能体 1：研究员（Researcher）\nresearcher = Agent(\n    agent_name=\"Researcher\",\n    system_prompt=\"你的工作是研究提供的主题，并提供详细的摘要。\",\n    model_name=\"gpt-5.4\",\n)\n\n# 智能体 2：写手（Writer）\nwriter = Agent(\n    agent_name=\"Writer\",\n    system_prompt=\"你的工作是根据研究摘要撰写一篇优美、引人入胜的博客文章。\",\n    model_name=\"gpt-5.4\",\n)\n\n# 创建一个顺序工作流，研究员的输出作为写手的输入\nworkflow = SequentialWorkflow(agents=[researcher, writer])\n\n# 在任务上运行工作流\nfinal_post = workflow.run(\"人工智能的历史与未来\")\nprint(final_post)\n```\n\n-----\n\n\n### ConcurrentWorkflow\n\n`ConcurrentWorkflow` 同时运行多个智能体，允许任务并行执行。该架构大幅缩短了可并行任务的执行时间，非常适合高吞吐量场景，其中多个智能体同时处理相似任务。\n\n```python\nfrom swarms import Agent, ConcurrentWorkflow\n\n# 为不同分析任务创建智能体\nmarket_analyst = Agent(\n    agent_name=\"Market-Analyst\",\n    system_prompt=\"分析市场趋势，并就给定主题提供洞察。\",\n    model_name=\"gpt-5.4\",\n    max_loops=1,\n)\n\nfinancial_analyst = Agent(\n    agent_name=\"Financial-Analyst\", \n    system_prompt=\"就给定主题提供财务分析和建议。\",\n    model_name=\"gpt-5.4\",\n    max_loops=1,\n)\n\nrisk_analyst = Agent(\n    agent_name=\"Risk-Analyst\",\n    system_prompt=\"评估风险，并就给定主题提供风险管理策略。\",\n    model_name=\"gpt-5.4\", \n    max_loops=1,\n)\n\n# 创建并发工作流\nconcurrent_workflow = ConcurrentWorkflow(\n    agents=[market_analyst, financial_analyst, risk_analyst],\n    max_loops=1,\n)\n\n# 同时在相同任务上运行所有智能体\nresults = concurrent_workflow.run(\n    \"分析人工智能技术对医疗行业潜在影响\"\n)\n\nprint(results)\n```\n\n### AgentRearrange\n\n受 `einsum` 启发，`AgentRearrange` 允许你使用简单的基于字符串的语法定义智能体（agent）之间复杂的非线性关系。[了解更多](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fagent_rearrange\u002F)。该架构非常适合编排动态工作流，其中智能体可以并行、顺序运行，或按你选择的任意组合方式协同工作。\n\n```python\nfrom swarms import Agent, AgentRearrange\n\n# 定义智能体\nresearcher = Agent(agent_name=\"researcher\", model_name=\"gpt-5.4\")\nwriter = Agent(agent_name=\"writer\", model_name=\"gpt-5.4\")\neditor = Agent(agent_name=\"editor\", model_name=\"gpt-5.4\")\n\n# 定义流程：researcher 同时将任务发送给 writer 和 editor\n# 这是一个一对多的关系\nflow = \"researcher -> writer, editor\"\n\n# 创建重排系统\nrearrange_system = AgentRearrange(\n    agents=[researcher, writer, editor],\n    flow=flow,\n)\n\n# 运行群组\noutputs = rearrange_system.run(\"Analyze the impact of AI on modern cinema.\")\nprint(outputs)\n```\n\n\n\u003C!-- \n### GraphWorkflow\n\n`GraphWorkflow` 使用有向无环图（Directed Acyclic Graph, DAG）来编排任务，允许你管理复杂依赖关系，其中某些任务必须等待其他任务完成后才能执行。\n\n**描述：** 对于构建复杂流水线（如软件开发或复杂项目管理）至关重要，其中任务顺序和依赖关系非常关键。\n\n```python\nfrom swarms import Agent, GraphWorkflow, Node, Edge, NodeType\n\n# 定义智能体和一个简单的 Python 函数作为节点\ncode_generator = Agent(agent_name=\"CodeGenerator\", system_prompt=\"Write Python code for the given task.\", model_name=\"gpt-5.4\")\ncode_tester = Agent(agent_name=\"CodeTester\", system_prompt=\"Test the given Python code and find bugs.\", model_name=\"gpt-5.4\")\n\n# 为图创建节点\nnode1 = Node(id=\"generator\", agent=code_generator)\nnode2 = Node(id=\"tester\", agent=code_tester)\n\n# 创建图并定义依赖关系\ngraph = GraphWorkflow()\ngraph.add_nodes([node1, node2])\ngraph.add_edge(Edge(source=\"generator\", target=\"tester\")) # Tester 在 generator 之后运行\n\n# 设置入口点和出口点\ngraph.set_entry_points([\"generator\"])\ngraph.set_end_points([\"tester\"])\n\n# 运行图工作流\nresults = graph.run(\"Create a function that calculates the factorial of a number.\")\nprint(results)\n``` -->\n\n----\n\n### SwarmRouter：通用群组编排器\n\n`SwarmRouter` 通过提供一个统一接口来运行任意类型的群组（swarm），从而简化复杂工作流的构建。你无需导入和管理不同的群组类，只需更改 `swarm_type` 参数即可动态选择所需类型。[阅读完整文档](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fswarm_router\u002F)\n\n这使你的代码更简洁、更灵活，能够轻松在不同的多智能体协作策略之间切换。以下是一个完整示例，展示了如何定义智能体，然后使用 `SwarmRouter` 以不同的协作策略执行相同的任务。\n\n```python\nfrom swarms import Agent\nfrom swarms.structs.swarm_router import SwarmRouter, SwarmType\n\n# 定义几个通用智能体\nwriter = Agent(agent_name=\"Writer\", system_prompt=\"You are a creative writer.\", model_name=\"gpt-5.4\")\neditor = Agent(agent_name=\"Editor\", system_prompt=\"You are an expert editor for stories.\", model_name=\"gpt-5.4\")\nreviewer = Agent(agent_name=\"Reviewer\", system_prompt=\"You are a final reviewer who gives a score.\", model_name=\"gpt-5.4\")\n\n# 所有示例使用相同的智能体和任务\nagents = [writer, editor, reviewer]\ntask = \"Write a short story about a robot who discovers music.\"\n\n# --- 示例 1：SequentialWorkflow ---\n# 智能体按链式顺序依次运行：Writer -> Editor -> Reviewer。\nprint(\"Running a Sequential Workflow...\")\nsequential_router = SwarmRouter(swarm_type=SwarmType.SequentialWorkflow, agents=agents)\nsequential_output = sequential_router.run(task)\nprint(f\"Final Sequential Output:\\n{sequential_output}\\n\")\n\n# --- 示例 2：ConcurrentWorkflow ---\n# 所有智能体同时接收相同的初始任务并并行运行。\nprint(\"Running a Concurrent Workflow...\")\nconcurrent_router = SwarmRouter(swarm_type=SwarmType.ConcurrentWorkflow, agents=agents)\nconcurrent_outputs = concurrent_router.run(task)\n# 返回每个智能体输出的字典\nfor agent_name, output in concurrent_outputs.items():\n    print(f\"Output from {agent_name}:\\n{output}\\n\")\n\n# --- 示例 3：MixtureOfAgents ---\n# 所有智能体并行运行，一个特殊的“聚合器”（aggregator）智能体综合它们的输出。\nprint(\"Running a Mixture of Agents Workflow...\")\naggregator = Agent(\n    agent_name=\"Aggregator\",\n    system_prompt=\"Combine the story, edits, and review into a final document.\",\n    model_name=\"gpt-5.4\"\n)\nmoa_router = SwarmRouter(\n    swarm_type=SwarmType.MixtureOfAgents,\n    agents=agents,\n    aggregator_agent=aggregator, # MoA 需要一个聚合器\n)\naggregated_output = moa_router.run(task)\nprint(f\"Final Aggregated Output:\\n{aggregated_output}\\n\")\n```\n\n\n`SwarmRouter` 是简化多智能体编排的强大工具。它提供了一种一致且灵活的方式来部署不同的协作策略，让你用更少的代码构建更复杂的应用程序。\n\n-------\n\n### MixtureOfAgents (MoA)\n\n`MixtureOfAgents` 架构通过将任务并行分发给多个“专家”智能体来处理任务。这些智能体产生的多样化输出随后由一个聚合器（aggregator）智能体进行综合，生成最终的高质量结果。[在此了解更多](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fmoa_example\u002F)\n\n```python\nfrom swarms import Agent, MixtureOfAgents\n\n# 定义专家智能体\nfinancial_analyst = Agent(agent_name=\"FinancialAnalyst\", system_prompt=\"Analyze financial data.\", model_name=\"gpt-5.4\")\nmarket_analyst = Agent(agent_name=\"MarketAnalyst\", system_prompt=\"Analyze market trends.\", model_name=\"gpt-5.4\")\nrisk_analyst = Agent(agent_name=\"RiskAnalyst\", system_prompt=\"Analyze investment risks.\", model_name=\"gpt-5.4\")\n\n# 定义聚合器智能体\naggregator = Agent(\n    agent_name=\"InvestmentAdvisor\",\n    system_prompt=\"Synthesize the financial, market, and risk analyses to provide a final investment recommendation.\",\n    model_name=\"gpt-5.4\"\n)\n\n# 创建 MoA 群组\nmoa_swarm = MixtureOfAgents(\n    agents=[financial_analyst, market_analyst, risk_analyst],\n    aggregator_agent=aggregator,\n)\n\n# 运行群组\nrecommendation = moa_swarm.run(\"Should we invest in NVIDIA stock right now?\")\nprint(recommendation)\n```\n\n----\n\n### GroupChat\n\n`GroupChat` 创建一个多智能体对话环境，让多个智能体可以互动、讨论并协作解决问题。你可以定义发言顺序，也可以让顺序动态决定。该架构非常适合那些能从辩论和多视角推理中受益的任务，例如合同谈判、头脑风暴或复杂决策。\n\n```python\nfrom swarms import Agent, GroupChat\n```\n\n# 定义用于辩论的智能体（Agent）\ntech_optimist = Agent(agent_name=\"TechOptimist\", system_prompt=\"论证人工智能对社会的益处。\", model_name=\"gpt-5.4\")\ntech_critic = Agent(agent_name=\"TechCritic\", system_prompt=\"反对人工智能的无节制发展。\", model_name=\"gpt-5.4\")\n\n# 创建群组聊天\nchat = GroupChat(\n    agents=[tech_optimist, tech_critic],\n    max_loops=4, # 限制对话轮次\n)\n\n# 使用初始话题运行聊天\nconversation_history = chat.run(\n    \"让我们讨论人工智能对社会的影响。\"\n)\n\n# 打印完整对话\nfor message in conversation_history:\n    print(f\"[{message['agent_name']}]: {message['content']}\")\n```\n\n----\n\n### HierarchicalSwarm（分层智能体群）\n\n`HierarchicalSwarm` 实现了“主管-员工”（director-worker）模式：一个中央主管智能体（director agent）制定全面计划，并将具体任务分配给专业化的员工智能体（worker agents）。主管会评估结果，并可在反馈循环中发布新指令，非常适合复杂项目管理和团队协作场景。\n\n```python\nfrom swarms import Agent, HierarchicalSwarm\n\n# 定义专业化员工智能体\ncontent_strategist = Agent(\n    agent_name=\"Content-Strategist\",\n    system_prompt=\"你是一位资深内容策略师。制定全面的内容策略、编辑日历和内容路线图。\",\n    model_name=\"gpt-5.4\"\n)\n\ncreative_director = Agent(\n    agent_name=\"Creative-Director\", \n    system_prompt=\"你是一位创意总监。制定引人注目的广告概念、视觉方向和营销活动创意。\",\n    model_name=\"gpt-5.4\"\n)\n\nseo_specialist = Agent(\n    agent_name=\"SEO-Specialist\",\n    system_prompt=\"你是一位 SEO 专家。进行关键词研究、内容优化，并制定自然增长策略。\",\n    model_name=\"gpt-5.4\"\n)\n\nbrand_strategist = Agent(\n    agent_name=\"Brand-Strategist\",\n    system_prompt=\"你是一位品牌策略师。制定品牌定位、识别系统和市场差异化策略。\",\n    model_name=\"gpt-5.4\"\n)\n\n# 创建带主管的分层智能体群\nmarketing_swarm = HierarchicalSwarm(\n    name=\"Marketing-Team-Swarm\",\n    description=\"由主管协调的专业化营销团队\",\n    agents=[content_strategist, creative_director, seo_specialist, brand_strategist],\n    max_loops=2,  # 允许反馈与优化\n    verbose=True\n)\n\n# 在复杂的营销挑战上运行智能体群\nresult = marketing_swarm.run(\n    \"为一款新的 SaaS 产品发布制定全面的营销策略。\"\n    \"该产品是一款面向中小企业的项目管理工具。\"\n    \"协调团队共同制定内容策略、创意活动、SEO 优化和品牌定位，确保各部分协同一致。\"\n)\n\nprint(result)\n```\n\n`HierarchicalSwarm` 擅长以下方面：\n- **复杂项目管理**：将大型任务分解为专业化的子任务\n- **团队协作**：确保所有智能体朝着统一目标工作\n- **质量控制**：主管提供反馈和优化循环\n- **可扩展的工作流**：可根据需要轻松添加新的专业化智能体\n\n---\n\n### HeavySwarm（重型智能体群）\n\n`HeavySwarm` 实现了一个受 X.AI 的 Grok heavy 启发的五阶段工作流。它使用专业化智能体（研究、分析、替代方案、验证）通过智能问题生成、并行执行和综合分析，提供全面的任务分析。该架构在需要深入调研和多角度视角的复杂研究与分析任务中表现卓越。\n\n```python\nfrom swarms import HeavySwarm\n\n# Pip install swarms-tools\nfrom swarms_tools import exa_search\n\nswarm = HeavySwarm(\n    name=\"Gold ETF Research Team\",\n    description=\"一支研究最佳黄金 ETF 的智能体团队\",\n    worker_model_name=\"claude-sonnet-4-20250514\",\n    show_dashboard=True,\n    question_agent_model_name=\"gpt-4.1\",\n    loops_per_agent=1,\n    agent_prints_on=False,\n    worker_tools=[exa_search],\n    random_loops_per_agent=True,\n)\n\nprompt = (\n    \"找出最佳的 3 只黄金 ETF。对于每只 ETF，请提供其股票代码、\"\n    \"全称、当前价格、费用比率、资产管理规模，以及\"\n    \"简要说明其为何被视为最佳之一。请以清晰、结构化的格式呈现信息，适合投资者阅读。从网络抓取数据。\"\n)\n\nout = swarm.run(prompt)\nprint(out)\n\n```\n\n`HeavySwarm` 提供以下能力：\n\n- **五阶段分析**：问题生成、研究、分析、替代方案和验证\n\n- **专业化智能体**：每个阶段使用专门构建的智能体以获得最佳结果\n\n- **全面覆盖**：多角度视角与深入调研\n\n- **实时仪表盘**：可选的分析过程可视化\n\n- **结构化输出**：组织良好且可操作的结果\n\n该架构非常适合财务分析、战略规划、研究报告，以及任何需要深入、多维度分析的任务。[了解更多关于 HeavySwarm 的信息](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fheavy_swarm\u002F)\n\n---\n\n### MAKER\n\n`MAKER` 实现了**最大化智能体分解**（maximal agentic decomposition），结合**领先 k 票胜出机制**（first-to-ahead-by-k voting）和**红旗标记**（red-flagging）：你提供 `format_prompt`、`parse_response`，以及可选的 `validate_response` \u002F `update_state`，然后运行固定步数（或直到满足停止条件）。每一步都会启动一个专注的单次执行智能体（one-shot `Agent`）（或循环使用你提供的智能体池），直到某个解析后的答案领先其他答案达到 `k` 票。这与论文《[Solving a Million-Step LLM Task with Zero Errors](https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.09030)》中的错误纠正机制一致。[完整文档](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fmaker\u002F)\n\n```python\nfrom swarms.structs.maker import MAKER\n\nmaker = MAKER(\n    model_name=\"gpt-4.1-mini\",\n    system_prompt=\"你以每步一行清晰语句的方式完成任务。\",\n    k=3,\n)\n\n# 可选：为你的领域重写 format_prompt \u002F parse_response \u002F validate_response。\nresults = maker.run(\n    task=\"列出类型化 API 的三个简洁优势，每步一条。\",\n    max_steps=3,\n)\nprint(results)\n```\n\n当 `k` 值较大时，为降低延迟，可使用 `run_parallel_voting`，传入相同的 `task` 和 `max_steps`。\n\n---\n\n### 社交算法（Social Algorithms）\n\n**社交算法**（Social Algorithms）提供了一个灵活的框架，用于定义智能体之间的自定义通信模式。你可以上传任意社交算法作为可调用对象（callable），定义通信顺序，使智能体能够以复杂的方式相互交流。[了解更多关于社交算法的信息](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Fsocial_algorithms\u002F)\n\n```python\nfrom swarms import Agent, SocialAlgorithms\n```\n\n# 定义一个自定义的社交算法（social algorithm）\ndef research_analysis_synthesis_algorithm(agents, task, **kwargs):\n    # Agent 1 对主题进行研究\n    research_result = agents[0].run(f\"Research: {task}\")\n    \n    # Agent 2 分析研究结果\n    analysis = agents[1].run(f\"Analyze this research: {research_result}\")\n    \n    # Agent 3 综合（synthesize）研究发现\n    synthesis = agents[2].run(f\"Synthesize: {research_result} + {analysis}\")\n    \n    return {\n        \"research\": research_result,\n        \"analysis\": analysis,\n        \"synthesis\": synthesis\n    }\n\n# 创建智能体（Agent）\nresearcher = Agent(\n  agent_name=\"Researcher\",\n  agent_description=\"Expert in comprehensive research and information gathering.\",\n  model_name=\"gpt-4.1\"\n)\nanalyst = Agent(\n  agent_name=\"Analyst\",\n  agent_description=\"Specialist in analyzing and interpreting data.\",\n  model_name=\"gpt-4.1\"\n)\nsynthesizer = Agent(\n  agent_name=\"Synthesizer\",\n  agent_description=\"Focused on synthesizing and integrating research insights.\",\n  model_name=\"gpt-4.1\"\n)\n\n# 创建社交算法\nsocial_alg = SocialAlgorithms(\n    name=\"Research-Analysis-Synthesis\",\n    agents=[researcher, analyst, synthesizer],\n    social_algorithm=research_analysis_synthesis_algorithm,\n    verbose=True\n)\n\n# 运行算法\nresult = social_alg.run(\"The impact of AI on healthcare\")\nprint(result.final_outputs)\n```\n\n非常适合用于实现复杂的多智能体工作流、协作式问题求解以及自定义通信协议。\n\n---\n\n### 智能体编排协议（Agent Orchestration Protocol, AOP）\n\n**智能体编排协议（AOP）** 是一个强大的框架，用于将智能体部署和管理为分布式服务。AOP 使智能体能够通过标准化协议被发现、管理和执行，非常适合构建可扩展的多智能体系统。[了解更多关于 AOP 的信息](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Faop\u002F)\n\n```python\nfrom swarms import Agent\nfrom swarms.structs.aop import AOP\n\n# 创建专用智能体\nresearch_agent = Agent(\n    agent_name=\"Research-Agent\",\n    agent_description=\"Expert in research and data collection\",\n    model_name=\"anthropic\u002Fclaude-sonnet-4-5\",\n    max_loops=1,\n    tags=[\"research\", \"data-collection\", \"analysis\"],\n    capabilities=[\"web-search\", \"data-gathering\", \"report-generation\"],\n    role=\"researcher\"\n)\n\nanalysis_agent = Agent(\n    agent_name=\"Analysis-Agent\", \n    agent_description=\"Expert in data analysis and insights\",\n    model_name=\"anthropic\u002Fclaude-sonnet-4-5\",\n    max_loops=1,\n    tags=[\"analysis\", \"data-processing\", \"insights\"],\n    capabilities=[\"statistical-analysis\", \"pattern-recognition\", \"visualization\"],\n    role=\"analyst\"\n)\n\n# 创建 AOP 服务器\ndeployer = AOP(\n    server_name=\"ResearchCluster\",\n    port=8000,\n    verbose=True\n)\n\n# 将智能体添加到服务器\ndeployer.add_agent(\n    agent=research_agent,\n    tool_name=\"research_tool\",\n    tool_description=\"Research and data collection tool\",\n    timeout=30,\n    max_retries=3\n)\n\ndeployer.add_agent(\n    agent=analysis_agent,\n    tool_name=\"analysis_tool\", \n    tool_description=\"Data analysis and insights tool\",\n    timeout=30,\n    max_retries=3\n)\n\n# 列出所有已注册的智能体\nprint(\"Registered agents:\", deployer.list_agents())\n\n# 启动 AOP 服务器\ndeployer.run()\n```\n\n非常适合部署大规模多智能体系统。[阅读完整的 AOP 文档](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Faop\u002F)\n\n---\n\n## 文档\n\n文档位于：[docs.swarms.world](https:\u002F\u002Fdocs.swarms.world)\n\n\n---\n\n## 示例\n\n探索全面的示例和教程，学习如何高效使用 Swarms。\n\n| 类别 | 示例 | 描述 | 链接 |\n|----------|---------|-------------|------|\n| **基础示例** | Basic Agent（基础代理） | 简单的代理设置与使用 | [Basic Agent](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fbasic_agent\u002F) |\n| **基础示例** | Agent with Tools（带工具的代理） | 使用各种工具的代理 | [Agent with Tools](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fagent_with_tools\u002F) |\n| **基础示例** | Agent with Structured Outputs（带结构化输出的代理） | 处理结构化数据输出 | [Structured Outputs](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fagent_structured_outputs\u002F) |\n| **基础示例** | Agent with MCP Integration（带 MCP 集成的代理） | Model Context Protocol（模型上下文协议）集成 | [MCP Integration](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fagent_with_mcp\u002F) |\n| **基础示例** | Vision Processing（视觉处理） | 具备图像处理能力的代理 | [Vision Processing](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fvision_processing\u002F) |\n| **基础示例** | Multiple Images（多图像处理） | 处理多张图像 | [Multiple Images](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fmultiple_images\u002F) |\n| **基础示例** | Vision and Tools（视觉与工具结合） | 将视觉能力与工具使用相结合 | [Vision and Tools](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fvision_tools\u002F) |\n| **基础示例** | Agent Streaming（代理流式输出） | 实时代理输出流 | [Agent Streaming](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fexamples\u002Fagent_stream\u002F) |\n| **基础示例** | Agent Output Types（代理输出类型） | 不同的输出格式与类型 | [Output Types](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fagent_output_types\u002F) |\n| **基础示例** | Gradio Chat Interface（Gradio 聊天界面） | 构建交互式聊天界面 | [Gradio UI](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fui\u002Fmain\u002F) |\n| **模型提供商** | Model Providers Overview（模型提供商概览） | 支持模型的完整指南 | [Model Providers](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fmodel_providers\u002F) |\n| **模型提供商** | OpenAI | OpenAI 模型集成 | [OpenAI Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fopenai_example\u002F) |\n| **模型提供商** | Anthropic | Claude 模型集成 | [Anthropic Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fclaude\u002F) |\n| **模型提供商** | Groq | Groq 模型集成 | [Groq Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fgroq\u002F) |\n| **模型提供商** | Cohere | Cohere 模型集成 | [Cohere Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fcohere\u002F) |\n| **模型提供商** | DeepSeek | DeepSeek 模型集成 | [DeepSeek Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fdeepseek\u002F) |\n| **模型提供商** | Ollama | 本地 Ollama 模型集成 | [Ollama Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Follama\u002F) |\n| **模型提供商** | OpenRouter | OpenRouter 模型集成 | [OpenRouter Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fopenrouter\u002F) |\n| **模型提供商** | XAI | XAI 模型集成 | [XAI Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fxai\u002F) |\n| **模型提供商** | Llama4 | Llama4 模型集成 | [Llama4 Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fllama4\u002F) |\n| **多代理架构** | HierarchicalSwarm（分层 Swarm） | 分层代理编排 | [HierarchicalSwarm Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fhierarchical_swarm_example\u002F) |\n| **多代理架构** | Hybrid Hierarchical-Cluster Swarm（混合分层-集群 Swarm） | 高级分层模式 | [HHCS Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fhhcs_examples\u002F) |\n| **多代理架构** | GroupChat（群聊） | 多代理对话 | [GroupChat Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fgroupchat_example\u002F) |\n| **多代理架构** | Sequential Workflow（顺序工作流） | 逐步执行的代理工作流 | [Sequential Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fsequential_example\u002F) |\n| **多代理架构** | SwarmRouter（Swarm 路由器） | 通用 Swarm 编排 | [SwarmRouter Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fswarm_router\u002F) |\n| **多代理架构** | MultiAgentRouter（多代理路由器） | 最简路由器示例 | [MultiAgentRouter Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fmulti_agent_router_minimal\u002F) |\n| **多代理架构** | ConcurrentWorkflow（并发工作流） | 并行代理执行 | [Concurrent Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fconcurrent_workflow\u002F) |\n| **多代理架构** | Mixture of Agents（代理混合体，MoA） | 专家代理协作 | [MoA Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fmoa_example\u002F) |\n| **多代理架构** | Unique Swarms（独特 Swarm） | 专用 Swarm 模式 | [Unique Swarms](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Funique_swarms\u002F) |\n| **多代理架构** | Agents as Tools（将代理作为工具） | 在工作流中将代理用作工具 | [Agents as Tools](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fagents_as_tools\u002F) |\n| **多代理架构** | Aggregate Responses（聚合响应） | 合并多个代理的输出 | [Aggregate Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Faggregate\u002F) |\n| **多代理架构** | Interactive GroupChat（交互式群聊） | 实时代理交互 | [Interactive GroupChat](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Figc_example\u002F) |\n| **部署方案** | Agent Orchestration Protocol (AOP)（代理编排协议） | 将代理作为分布式服务部署，支持发现与管理 | [AOP Reference](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fstructs\u002Faop\u002F) |\n| **应用案例** | Advanced Research System（高级研究系统） | 受 Anthropic 研究方法启发的多代理研究系统 | [AdvancedResearch](https:\u002F\u002Fgithub.com\u002FThe-Swarm-Corporation\u002FAdvancedResearch) |\n| **应用案例** | Hospital Simulation（医院模拟） | 基于多代理架构的医疗模拟系统 | [HospitalSim](https:\u002F\u002Fgithub.com\u002FThe-Swarm-Corporation\u002FHospitalSim) |\n| **应用案例** | Browser Agents（浏览器代理） | 使用代理进行网页自动化 | [Browser Agents](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fswarms_of_browser_agents\u002F) |\n| **应用案例** | Medical Analysis（医疗分析） | 医疗健康领域的应用 | [Medical Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fswarms_api_medical\u002F) |\n| **应用案例** | Finance Analysis（金融分析） | 金融领域的应用 | [Finance Examples](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fexamples\u002Fswarms_api_finance\u002F) |\n| **实战手册与模板** | Examples Overview（示例概览） | 完整的示例目录 | [Examples Index](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fexamples\u002F) |\n| **实战手册与模板** | Cookbook Index（实战手册索引） | 精选示例合集 | [Cookbook](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fexamples\u002Fcookbook_index\u002F) |\n| **实战手册与模板** | Paper Implementations（论文实现） | 研究论文的实现 | [Paper Implementations](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fexamples\u002Fpaper_implementations\u002F) |\n| **实战手册与模板** | Templates & Applications（模板与应用） | 可复用的模板 | [Templates](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fexamples\u002Ftemplates\u002F) |\n\n## 为 Swarms 做出贡献\n\n我们的使命是通过提供企业级、生产就绪（production-ready）的基础设施，加速向完全自主化的世界经济转型，从而实现数百万个自主智能体（autonomous agents）的无缝部署与编排。我们正在构建智能体经济的操作系统，而实现这一目标离不开您的帮助。\n\nSwarms 由社区共建，也为社区服务。我们相信，协作开发是推动多智能体 AI（multi-agent AI）能力边界的关键。您的贡献不仅受到欢迎——更是我们使命中不可或缺的一部分。[了解更多关于为何应为 Swarms 做贡献的原因](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fcontributors\u002Fmain\u002F)\n\n### 为什么要贡献？\n\n加入我们，您将有机会：\n\n* **站在智能体技术的前沿**：塑造自主智能体技术的未来，帮助构建一个生产级、开源的框架。\n\n* **加入充满活力的社区**：与一群热情且不断壮大的智能体开发者、研究人员和爱好者协作。\n\n* **产生切实影响**：无论您是在修复 bug、添加新功能，还是改进文档，您的工作都将被用于真实世界的应用中。\n\n* **学习与成长**：获得先进 AI 概念的实践经验，并提升您的软件工程技能。\n\n在我们的官方 [**贡献者指南**](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fcontributors\u002Fmain\u002F) 中，了解更多关于我们的使命以及成为贡献者的好处。\n\n### 如何开始贡献\n\n我们已尽可能简化贡献流程。以下是您可以参与的方式：\n\n1. **寻找可处理的问题**：最好的入门方式是访问我们的 [**贡献项目看板**](https:\u002F\u002Fgithub.com\u002Fusers\u002Fkyegomez\u002Fprojects\u002F1)。寻找带有 `good first issue` 标签的问题——这些问题是专门为新贡献者挑选的。\n\n2. **报告 Bug 或提出功能请求**：有新想法或发现某些功能异常？我们非常期待您的反馈。请在 GitHub Issues 页面上 [**提交 Bug 报告或功能请求**](https:\u002F\u002Fgithub.com\u002Fkyegomez\u002Fswarms\u002Fissues)。\n\n3. **了解我们的工作流程和标准**：在提交您的工作前，请务必阅读完整的 [**贡献指南**](https:\u002F\u002Fgithub.com\u002Fkyegomez\u002Fswarms\u002Fblob\u002Fmaster\u002FCONTRIBUTING.md)。为帮助维持代码质量，我们也鼓励您阅读 [**代码整洁性指南**](https:\u002F\u002Fdocs.swarms.world\u002Fen\u002Flatest\u002Fswarms\u002Fframework\u002Fcode_cleanliness\u002F)。\n\n4. **加入讨论**：如需参与路线图讨论或与其他开发者交流，请加入我们的 [**Discord 社区**](https:\u002F\u002Fdiscord.gg\u002FEamjgSaEQf)。\n\n### 感谢我们的贡献者\n\n感谢您为 Swarms 做出的贡献。您的工作备受赞赏并得到认可。\n\n\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fkyegomez\u002Fswarms\u002Fgraphs\u002Fcontributors\">\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fkyegomez_swarms_readme_4afdaf65db2f.png\" \u002F>\n\u003C\u002Fa>\n\n### 感谢我们的社区\n\n我们对每一位支持 Swarms 的朋友都心怀感激！您的 Star、Fork 和贡献让这个项目每天变得更好。\n\n[![Forkers repo roster for @kyegomez\u002Fswarms](https:\u002F\u002Freporoster.com\u002Fforks\u002Fkyegomez\u002Fswarms)](https:\u002F\u002Fgithub.com\u002Fkyegomez\u002Fswarms\u002Fnetwork\u002Fmembers)\n\n[![Stargazers repo roster for @kyegomez\u002Fswarms](https:\u002F\u002Freporoster.com\u002Fstars\u002Fkyegomez\u002Fswarms)](https:\u002F\u002Fgithub.com\u002Fkyegomez\u002Fswarms\u002Fstargazers)\n\n-----\n\n## 加入 Swarms 社区\n\n加入我们的智能体工程师与研究人员社区，获取技术支持、前沿动态，以及独家的世界级智能体工程洞见！\n\n| 平台 | 描述 | 链接 |\n|----------|-------------|------|\n| 文档 | 官方文档与指南 | [docs.swarms.world](https:\u002F\u002Fdocs.swarms.world) |\n| 博客 | 最新动态与技术文章 | [Medium](https:\u002F\u002Fmedium.com\u002F@kyeg) |\n| Discord | 实时聊天与社区支持 | [加入 Discord](https:\u002F\u002Fdiscord.gg\u002FEamjgSaEQf) |\n| Twitter | 最新新闻与公告 | [@swarms_corp](https:\u002F\u002Ftwitter.com\u002Fswarms_corp) |\n| LinkedIn | 职业网络与更新 | [The Swarm Corporation](https:\u002F\u002Fwww.linkedin.com\u002Fcompany\u002Fthe-swarm-corporation) |\n| YouTube | 教程与演示 | [Swarms 频道](https:\u002F\u002Fwww.youtube.com\u002Fchannel\u002FUC9yXyitkbU_WSy7bd_41SqQ) |\n| 活动 | 参加我们的社区活动 | [立即报名](https:\u002F\u002Flu.ma\u002Fswarms_calendar) |\n| 入门会议 | 与 Swarms 创始人兼首席维护者 Kye Gomez 一起完成入门 | [预约会议](https:\u002F\u002Fcal.com\u002Fswarms\u002Fswarms-onboarding-session) |\n\n------\n\n## 引用\n\n如果您在研究中使用了 **swarms**，请通过引用 [CITATION.cff](.\u002FCITATION.cff) 中的元数据来引用本项目。\n\n```bibtex\n@misc{SWARMS_2022,\n  author  = {Kye Gomez and Pliny and Zack Bradshaw and Ilumn and Harshal and the Swarms Community},\n  title   = {{Swarms: Production-Grade Multi-Agent Infrastructure Platform}},\n  year    = {2022},\n  howpublished = {\\url{https:\u002F\u002Fgithub.com\u002Fkyegomez\u002Fswarms}},\n  note    = {Documentation available at \\url{https:\u002F\u002Fdocs.swarms.world}},\n  version = {latest}\n```\n\n---\n\n# 许可证\n\nSwarms 采用 Apache License 2.0 许可证。[在此了解更多](.\u002FLICENSE)","# Swarms 快速上手指南\n\n## 环境准备\n\n- **Python 版本**：3.9 或更高版本\n- **操作系统**：Linux \u002F macOS \u002F Windows 均支持\n- **API 密钥（可选但推荐）**：\n  - `OPENAI_API_KEY`（使用 OpenAI 模型时必需）\n  - `ANTHROPIC_API_KEY`（使用 Claude 模型时必需）\n  - `GROQ_API_KEY`（使用 Groq 推理加速时可选）\n\n建议设置以下环境变量（可通过 `.env` 文件或系统环境变量配置）：\n\n```bash\nOPENAI_API_KEY=\"your-api-key\"\nWORKSPACE_DIR=\"agent_workspace\"\nANTHROPIC_API_KEY=\"your-anthropic-key\"\nGROQ_API_KEY=\"your-groq-key\"\n```\n\n> 💡 国内用户建议使用 [uv](https:\u002F\u002Fgithub.com\u002Fastral-sh\u002Fuv) 安装工具以获得更快的依赖解析速度，并可配合清华源等镜像加速安装。\n\n## 安装步骤\n\n### 推荐方式（使用 uv）\n\n```bash\n$ uv pip install swarms\n```\n\n### 使用 pip（国内可加 `-i https:\u002F\u002Fpypi.tuna.tsinghua.edu.cn\u002Fsimple` 加速）\n\n```bash\n$ pip3 install -U swarms\n```\n\n### 使用 Poetry\n\n```bash\n$ poetry add swarms\n```\n\n### 从源码安装（适用于开发贡献）\n\n```bash\n$ git clone https:\u002F\u002Fgithub.com\u002Fkyegomez\u002Fswarms.git\n$ cd swarms\n$ pip install -r requirements.txt\n```\n\n## 基本使用\n\n### 创建单个智能体（Agent）\n\n```python\nfrom swarms import Agent\n\n# 初始化一个智能体\nagent = Agent(\n    model_name=\"gpt-4\",        # 指定使用的 LLM（需有对应 API 密钥）\n    max_loops=\"auto\",          # 自动决定循环次数\n    interactive=True,          # 启用交互模式\n)\n\n# 执行任务\nresult = agent.run(\"多智能体系统的主要优势是什么？\")\nprint(result)\n```\n\n### 构建多智能体协作流程（Swarm）\n\n```python\nfrom swarms import Agent, SequentialWorkflow\n\n# 研究员智能体\nresearcher = Agent(\n    agent_name=\"Researcher\",\n    system_prompt=\"你的任务是深入研究指定主题并提供详细摘要。\",\n    model_name=\"gpt-4\",\n)\n\n# 写作智能体\nwriter = Agent(\n    agent_name=\"Writer\",\n    system_prompt=\"你的任务是根据研究员提供的摘要，撰写一篇生动、专业的博客文章。\",\n    model_name=\"gpt-4\",\n)\n\n# 创建顺序工作流\nworkflow = SequentialWorkflow(agents=[researcher, writer])\n\n# 执行端到端任务\nfinal_output = workflow.run(\"人工智能的历史与未来发展趋势\")\nprint(final_output)\n```\n\n> ✅ 提示：首次运行时请确保已正确配置对应模型的 API 密钥。更多高级功能（如自动构建 Swarm、集成 MCP\u002FX402 协议等）请参考 [官方文档](https:\u002F\u002Fdocs.swarms.world)。","某金融科技公司需要构建一个智能投研系统，由多个 AI 代理协同完成市场数据抓取、舆情分析、风险评估和投资建议生成等任务。\n\n### 没有 swarms 时\n- 多个 AI 代理各自独立运行，缺乏统一调度机制，任务执行顺序混乱，经常出现数据依赖错位。\n- 开发团队需手动编写大量胶水代码来串联不同模型（如 GPT、Claude、本地 LLM），集成成本高且难以维护。\n- 系统无法动态调整代理数量或角色，面对突发市场事件时响应迟缓，扩展性差。\n- 缺乏统一的日志、监控和错误回溯能力，线上问题排查耗时数小时甚至更久。\n- 与公司现有风控和合规系统对接困难，每次新增功能都要重写适配层。\n\n### 使用 swarms 后\n- 通过 swarms 的图式代理网络和工作流编排能力，清晰定义各代理间的依赖关系，确保数据流和逻辑流严格按业务规则执行。\n- 利用 swarms 内置的多模型支持和标准化接口，快速接入不同大模型，无需重复开发适配逻辑，开发效率提升 60% 以上。\n- 借助动态代理组合与自动扩缩容机制，系统可在市场波动剧烈时自动增派舆情分析代理，实时响应能力显著增强。\n- swarms 提供的可观测性工具链（日志、指标、追踪）让运维团队能在分钟级定位并修复异常。\n- 通过 swarms 的企业级集成能力，无缝对接内部合规引擎和数据库，确保所有输出符合金融监管要求。\n\nswarms 将原本碎片化的多代理系统整合为一个高可用、可治理、易扩展的智能协作平台，真正实现了企业级 AI 工作流的工业化落地。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fkyegomez_swarms_65489219.png","kyegomez","Kye Gomez","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fkyegomez_0b95c3bb.jpg","Founder of swarms.ai","Swarms","Palo Alto",null,"KyeGomezB","https:\u002F\u002Fgithub.com\u002Fkyegomez\u002Fswarms","https:\u002F\u002Fgithub.com\u002Fkyegomez",[86,90,94,98],{"name":87,"color":88,"percentage":89},"Python","#3572A5",99.5,{"name":91,"color":92,"percentage":93},"Shell","#89e051",0.3,{"name":95,"color":96,"percentage":97},"PowerShell","#012456",0.1,{"name":99,"color":100,"percentage":101},"Dockerfile","#384d54",0,6195,798,"2026-04-05T19:13:19","Apache-2.0","Linux, macOS, Windows","未说明",{"notes":109,"python":107,"dependencies":110},"README 中未明确说明操作系统、GPU、内存、Python 版本及具体依赖库要求；建议参考官方文档或源码中的 requirements.txt 获取详细依赖信息。支持通过 pip、uv、poetry 或源码安装。需要配置如 OPENAI_API_KEY 等环境变量以使用大模型功能。",[],[15,14,26,13],[113,114,115,116,67,117,118,119,120,121,122,123,124,125,126,127,128,129],"artificial-intelligence","gpt4","langchain","machine-learning","agents","prompt-engineering","prompt-toolkit","prompting","tree-of-thoughts","ai","chatgpt","gpt4all","huggingface","langchain-python","agentic-workflow","multi-agent-systems","agentic-ai",8,"2026-03-27T02:49:30.150509","2026-04-06T07:13:47.448251",[134,139,144,149,154,159],{"id":135,"question_zh":136,"answer_zh":137,"source_url":138},314,"运行示例脚本时失败，如何解决？","Swarms 要求使用 Python 3.10。请创建一个基于 Python 3.10 的虚拟环境，并确保已正确安装依赖。可尝试升级 Swarms：`pip3 install -U swarms`。如果问题仍然存在，请检查是否满足所有环境要求。","https:\u002F\u002Fgithub.com\u002Fkyegomez\u002Fswarms\u002Fissues\u002F417",{"id":140,"question_zh":141,"answer_zh":142,"source_url":143},315,"在使用 Azure OpenAI（如 o4-mini、o3）时出现参数不支持错误怎么办？","Azure OpenAI 不支持某些参数（如 top_p）。解决方法是在代码中全局设置 `litellm.drop_params = True`，例如：\n```python\nimport litellm\nlitellm.drop_params = True\nagent = Agent(model_name=\"azure\u002Fo4-mini\")\n```\n同时确保设置了正确的环境变量：`AZURE_API_BASE` 和 `AZURE_API_KEY`。","https:\u002F\u002Fgithub.com\u002Fkyegomez\u002Fswarms\u002Fissues\u002F842",{"id":145,"question_zh":146,"answer_zh":147,"source_url":148},316,"在 Windows 11 上安装 Swarms 失败怎么办？","请尝试更新 Swarms 到最新版本：`pip3 install -U swarms`。维护者已在新版本中修复了 Windows 安装问题。如果仍失败，请确认系统已安装必要的编译工具（如 Visual Studio Build Tools）。","https:\u002F\u002Fgithub.com\u002Fkyegomez\u002Fswarms\u002Fissues\u002F760",{"id":150,"question_zh":151,"answer_zh":152,"source_url":153},317,"导入 swarms 时出现 'TypeError: 'type' object is not subscriptable' 错误怎么办？","该错误通常是因为 Python 版本过低。Swarms 要求 Python 3.9 或更高版本。请升级 Python 至 3.9+。此外，如果使用 pydantic，可尝试降级到兼容版本：`pip3 install pydantic==1.10.12`。","https:\u002F\u002Fgithub.com\u002Fkyegomez\u002Fswarms\u002Fissues\u002F241",{"id":155,"question_zh":156,"answer_zh":157,"source_url":158},318,"使用 Ollama 本地模型时输出包含大量非响应信息（如 duration、eval_count 等）怎么办？","此问题是由于 Ollama 模型返回的原始响应未被正确解析。该问题已在 v7.8.4 后的版本中修复（参见 PR #847）。请升级 Swarms 到最新版：`pip3 install -U swarms`，以确保只输出纯净的模型响应内容。","https:\u002F\u002Fgithub.com\u002Fkyegomez\u002Fswarms\u002Fissues\u002F771",{"id":160,"question_zh":161,"answer_zh":162,"source_url":163},319,"运行 agent 时提示 'Agent' object has no attribute 'get_llm_init_params' 怎么办？","该错误通常出现在旧版本的 Swarms 中。请确保使用的是最新版本：`pip3 install -U swarms`。如果在 Google Colab 中运行，请重启运行时并重新安装最新版，以避免缓存旧代码导致的问题。","https:\u002F\u002Fgithub.com\u002Fkyegomez\u002Fswarms\u002Fissues\u002F370",[165,169,174,179,184,189,194,199,204,209,214,219,224,228,232,236,240,244,248,253],{"id":166,"version":167,"summary_zh":167,"released_at":168},109690,"6.8.1","2024-12-27T20:30:59",{"id":170,"version":171,"summary_zh":172,"released_at":173},109691,"5.3.7","- [GRAPH WORKFLOW]","2024-07-15T04:16:57",{"id":175,"version":176,"summary_zh":177,"released_at":178},109692,"2.5.0","Swarms Version 2.5.0 Changelog\r\n\r\nCommits on Dec 1, 2023\r\n\r\nFix: execute_futures_dict\r\n\r\nCommit ID: 452bf7d\r\nFix: execute_futures_dict\r\n\r\nCommit ID: 1c280b1\r\nCleanup\r\n\r\nCommit ID: 64ecedb\r\nAbstractLLM\r\n\r\nCommit ID: a0a0128\r\nUpdate base_multimodal_model.py\r\n\r\nCommit ID: 061db36\r\nFeature: Readme\r\n\r\nCommit ID: b726f04\r\nFeature: Docs for AbstractLLM, Docs for BaseMultiModalModel\r\n\r\nCommit ID: a7a6d54\r\nFeature: AutoScaler Prototype, Tests\r\n\r\nCommit ID: 3ae305e\r\nFeature: Agent prompt cleanup\r\n\r\nCommit ID: 84a8449\r\nCommits on Nov 30, 2023\r\n\r\nCleanup\r\n\r\nCommit ID: 0555a2b\r\nAbstractLLM\r\n\r\nCommit ID: d8fb6f8\r\nUpdate stable_diffusion.py\r\n\r\nCommit ID: e09001f\r\nCreate idea2img.py\r\n\r\nCommit ID: 8213769\r\nCreate idea2img.py\r\n\r\nCommit ID: b382d6b\r\nCreate idea2img.py\r\n\r\nCommit ID: 9eabc91\r\nCleanup\r\n\r\nCommit ID: a8449d2\r\nFeature: AbstractLLM\r\n\r\nCommit ID: a37b314\r\nCode Quality\r\n\r\nCommit ID: ec51149\r\nReadme\r\n\r\nCommit ID: 4e47528\r\nFeature: SequentialWorkflow\r\n\r\nCommit ID: 6e42d54\r\nFeature: Scripts for pytest, SequentialWorkflow\r\n\r\nCommit ID: a47f3a7\r\nCommits on Nov 29, 2023\r\n\r\nCleanup: Multi agent debate\r\n\r\nCommit ID: 2b6b050\r\nCleanup: Tests cleanup\r\n\r\nCommit ID: 4275efa\r\nCleanup\r\n\r\nCommit ID: b1bf3ef\r\nCleanup\r\n\r\nCommit ID: d08a2cb\r\nCleanup: Non Existent tests\r\n\r\nCommit ID: 45c6b11\r\nFeature: SequentialWorkflow\r\n\r\nCommit ID: a679d01\r\nCreate logistics.py\r\n\r\nCommit ID: 5de20b3\r\nUpload files\r\n\r\nCommit ID: 0d8f657\r\nCreate logistics.py\r\n\r\nCommit ID: b027c22\r\nCreate logistics.py\r\n\r\nCommit ID: dacc472\r\nUpdate pip\r\n\r\nCommit ID: 0ff1aa0\r\nCode Quality\r\n\r\nCommit ID: dc55006\r\nMerge pull request: Browse the repository\r\n\r\nCommit ID: 87e8b90","2023-12-01T22:40:34",{"id":180,"version":181,"summary_zh":182,"released_at":183},109693,"2.4.2","# 2.4.2 Release\r\n\r\n75f3cf1: [FEAT][Flow.run() img = None for conditional img inputs, BaseMultiMod...\r\na92a6a5: FEAT: [BEAUTIFY in GPT4Vision][Disable logging in init of swarms]\r\nf895497: Jarvis demo, base multimmodalmodel, whisperx -> whisperx_model\r\n51c82cf: Commits on Nov 24, 2023 gpt4vision features and api\r\n9390efb: cleanup\r\nb9fea7b: gpt4v docs\r\n66d9f70: GPT4Vision + multimodal\r\ndfa4197: Merge pull request #190 from elder-plinius\u002Fmaster\r\n539bfea: Add files via upload\r\n97a79ac: Create assembly.py\r\na4348a8: clean up outputs of multi modal autonomous agents\r\n9e6c427: tests for gpt4visionapi\r\n0802091: gpt4vision api\r\n9c3a292: dockerfile\r\n399099e: Merge pull request #181 from evelynmitchell\u002Fmaster\r\n55a1dfa: nougat fix\r\nee1ac00: clean up\r\nd88a31a: positive_med demo;\r\nb554d17: Merge pull request #184 from elder-plinius\u002Fmaster\r\n179ca14: Merge branch 'kyegomez:master' into master\r\n673b8ac: prompt name fixes for positive med\r\n29b6c9b: Update autotemp_example.py\r\nad3e8e7: Update and rename blog_gen_example.py to blog_gen_example.py\r\n66b9ba0: Rename blog_gen to blog_gen.py\r\ncd82fa6: Update and rename blog_gen.py to blog_gen\r\n14570d1: positive med\r\ned071e0: renamed tests to allow for pytest autodiscovery\r\na229685: black formatting\r\nd1144ec: Commits on Nov 23, 2023 requirements.txt clean up\r\n36a641b: requirements.txt\r\n9f20592: code quality fixes: line length = 80\r\n49c7b97: torch verisoning\r\nd97de1c: multi agent docs + playground code quality","2023-11-25T22:27:54",{"id":185,"version":186,"summary_zh":187,"released_at":188},109694,"2.3.1","Changelog for version 2.3.1:\r\n\r\nCommit on Nov 14, 2023:\r\nkyegomez committed b2dd2ef - tests for yi, stable diffusion, timm models, etc\r\nkyegomez committed dfea671 - no api key\r\nkyegomez committed 720a728 - mistral caller, openai version 2.8, llama function caller, tests for ...\r\nkyegomez committed 699c943 - mistral caller, openai version 2.8, llama function caller, tests for ...\r\nkyegomez committed 5f56023\r\nCommits on Nov 13, 2023:\r\nkyegomez committed 7db6930 - clean\r\nkyegomez committed 0d0e5e5 - Merge pull request #129 from kyegomez\u002Fdependabot\u002Fgithub_actions\u002Fcodac...\r\nkyegomez committed 694100e - Merge pull request #131 from kyegomez\u002Fdependabot\u002Fgithub_actions\u002Factio...\r\nkyegomez committed 39ba632 - Merge pull request #130 from kyegomez\u002Fdependabot\u002Fgithub_actions\u002Fsigst...\r\nkyegomez committed 2b05329 - Merge pull request #132 from kyegomez\u002Fdependabot\u002Fgithub_actions\u002Factio...\r\nkyegomez committed b2b9665 - Bump actions\u002Fcheckout from 3 to 4\r\ndependabot[bot] committed 1203d84 - Bump actions\u002Fsetup-python from 3 to 4\r\ndependabot[bot] committed 1cd0779 - Bump sigstore\u002Fcosign-installer from 3.1.1 to 3.2.0\r\ndependabot[bot] committed 1046e8a - Bump codacy\u002Fcodacy-analysis-cli-action from 1.1.0 to 4.3.0","2023-11-15T02:07:07",{"id":190,"version":191,"summary_zh":192,"released_at":193},109695,"2.2.2","- Removed fuyu in the __init__","2023-11-11T23:22:22",{"id":195,"version":196,"summary_zh":197,"released_at":198},109696,"2.2.1","[fuyu attempted fix](https:\u002F\u002Fgithub.com\u002Fkyegomez\u002Fswarms\u002Fcommit\u002Fbcf2b888739eaeac8747ec30eec06cb8c04e9eeb)","2023-11-11T22:49:14",{"id":200,"version":201,"summary_zh":202,"released_at":203},109697,"2.1.9","# Changelog for version 2.1.9 - Commits on Nov 11, 2023\r\n\r\nCLEANUP: swarms.agents, removed unused files [07bcd22]\r\n50+ tests for cohere [bdc7337]\r\nTests clean up, import file paths, and env PSG connection strings [4197920]\r\nEnvironment setup [0c6daf7]\r\nTests for groupchats, anthropic [9402dab]\r\nNew version [76fd9a0]\r\nTests, workflow fixes, torch version [2e6efb4]\r\nRelease workflow [e8e024f]\r\nCreated codacy.yml [d2b5fc9]\r\nCreated codeql.yml [6dcaab6]\r\nCreated docker-publish.yml [1d01402]\r\nCreated python-app.yml [aae6c03]\r\nCreated docker-image.yml [fdcfa0c]\r\nTools for flow and general cleanup [0e335b1]\r\nCommits on Nov 10, 2023\r\n\r\n2.1.7, no langchain experimental [a610ff7]\r\ndalle3 production grade ready [41e5f17]\r\nNo schemas [d26531f]\r\nCommits on Nov 9, 2023\r\n\r\nDockerfile [4596ddc]\r\nDockerfile running [991979d]\r\nDynamic max loops, GPT4 clean up [371da79]","2023-11-11T22:38:55",{"id":205,"version":206,"summary_zh":207,"released_at":208},109698,"2.1.7","Changelog:\r\n\r\n[2e6efb4] Release workflow\r\n[e8e024f] Create codacy.yml\r\n[d2b5fc9] Create codeql.yml\r\n[6dcaab6] Create docker-publish.yml\r\n[1d01402] Create python-app.yml\r\n[aae6c03] Create docker-image.yml\r\n[fdcfa0c] Tools for flow and general cleanup\r\n[0e335b1] Commits on Nov 10, 2023 - 2.1.7, no langchain experimental\r\n[a610ff7] Dalle3 production grade ready\r\n[41e5f17] No schemas\r\n[d26531f] Commits on Nov 9, 2023 - Dockerfile\r\n[4596ddc] Dockerfile running\r\n[991979d] Dynamic max loops, + gpt4 clean up\r\n[371da79] Account swarm + layout document fix\r\n[1dc1e8f] Commits on Nov 8, 2023 - Clean up\r\n[fe19f21] Yapf code quality\r\n[2e7905d] Code quality + new version + fuyu fixes\r\n[e0c712b] Update zephyr.py\r\n[73b549b] Fuyu + zephyr fixes\r\n[9aa4084] Anthropic + kosmo2 + fastvit\r\n[6b4c2d4] Anthropic working\r\n[f74a4da] Accountant swarm + autotemp agent\r\n[11b02a5] Commits on Nov 7, 2023 - Security\r\n[3d186fd] Create CODE_OF_CONDUCT.md\r\n[f16515a] Create SECURITY.md\r\n[570ed6a] 2.0.5\r\n[7a5e82b] Clean up unused code\r\n[bb496f4] Version\r\n[7bc6b9c] Docs for misral and groupchat\r\n[4f68c54] Merge pull request #100 from vyomakesh09\u002Fmaster","2023-11-11T15:14:20",{"id":210,"version":211,"summary_zh":212,"released_at":213},109699,"2.0.5","Changelog for Release 2.0.5:\r\n\r\n-   Nov 7, 2023:\r\n\r\n    -   Cleaned up unused code. [commit: 7a5e82b]\r\n    -   Removed useless code, no more worker, etc. [commit: bb496f4]\r\n    -   Added version. [commit: 7bc6b9c]\r\n    -   Added docs for Misral and Groupchat. [commit: 55a4e9a]\r\n    -   Merged pull request #100 from vyomakesh09\u002Fmaster. [commit: 4f68c54]\r\n    -   Fixed typo in bfloat16. [commit: b61fcf3]\r\n-   Nov 6, 2023:\r\n\r\n    -   Removed open interpreter, cleaned up docs, added add messages to flow. [commit: 16176e8]\r\n    -   Updated swarms. [commit: 62a4135]\r\n    -   Cleaned up code. [commit: 8dc9064]\r\n    -   Updated anthropic docs. [commit: 4c6bbad]\r\n    -   Fixed saved state error in dashboard. [commit: 97aa8bc]\r\n    -   Fixed playground and flow docs. [commit: 336bffe]","2023-11-07T21:39:49",{"id":215,"version":216,"summary_zh":217,"released_at":218},109700,"2.0.2","Changelog:\r\n\r\nCommits on Nov 6, 2023\r\n\r\nswarms (62a4135)\r\nclean up (8dc9064)\r\nanthropic docs (4c6bbad)\r\nsaved state in dashboard error (97aa8bc)\r\nplayground + flow docs fix (336bffe)\r\nDependencies clean up (a70a2b0)\r\ndalle3 (4fb38c1)\r\ndocs for gpt4v (c7d128d)\r\ndocs (b7aa21f)\r\nGPT4Vision + Dalle3 -> modules + tests + documentation (fd8919d)\r\nguides (fe48ec1)\r\nm (a97f759)\r\ndocs clean up -> corporate folder (881ec11)\r\nmulti modal auto agent + removed workflow.py (c520cda)\r\nswarms docs corporate + sequential workflow (6010c9d)\r\nsequential workflow (71da697)\r\nworkflow states (383412b)\r\nCommits on Nov 5, 2023\r\n\r\nsequential workflow docs (c94512d)\r\nsequential workflow tests, prototype with documentation (310230a)\r\nauto saved + fixed run method of flow (ba28f40)\r\nCommits on Nov 4, 2023\r\n\r\nanthropic tests (d4bd4fa)\r\nfuyu fix (1fb1932)\r\ndocs for DistilWhisperModel (6e6fe8d)\r\ntests for distilled whisperx (7e1d486)\r\ndistilled whisperx (75ebbe0)\r\nflow -> example.py (2f31a63)\r\ngroupchat (154f50c)\r\nCommits on Nov 3, 2023\r\n\r\nflow example, save and load state (7d888c6)\r\nflow (f53236a)\r\nexpanded permissions to allow welcome action run (4392b29)\r\nadded labeler.yml (f190d84)\r\nno stream (1b0cb87)\r\nflow (1162271)\r\nMerge pull request #61 from kyegomez\u002Fdependabot\u002Fgithub_actions\u002Faction... (17ff736)","2023-11-07T01:02:58",{"id":220,"version":221,"summary_zh":222,"released_at":223},109701,"1.3.2","- less dependencies","2023-08-09T21:56:58",{"id":225,"version":226,"summary_zh":222,"released_at":227},109702,"1.3.1","2023-08-09T21:53:59",{"id":229,"version":230,"summary_zh":81,"released_at":231},109703,"1.3.0","2023-08-09T21:36:58",{"id":233,"version":234,"summary_zh":81,"released_at":235},109704,"1.2.9","2023-08-09T21:30:57",{"id":237,"version":238,"summary_zh":238,"released_at":239},109705,"1.2.7","2023-08-09T21:25:38",{"id":241,"version":242,"summary_zh":81,"released_at":243},109706,"1.2.6","2023-08-09T20:55:40",{"id":245,"version":246,"summary_zh":246,"released_at":247},109707,"1.2.5","2023-08-09T20:41:51",{"id":249,"version":250,"summary_zh":251,"released_at":252},109708,"1.2.4","- verison","2023-08-09T17:04:58",{"id":254,"version":255,"summary_zh":81,"released_at":256},109709,"1.2.3","2023-08-09T16:55:04"]