[{"data":1,"prerenderedAt":2473},["ShallowReactive",2],{"header-counts":3,"footer-counts":6,"alt-main-crewai":8,"alt-list-crewai":618},{"tools":4,"reviews":5},98,27,{"tools":4,"reviews":5,"playbooks":7,"news":5},23,{"id":9,"title":10,"alternatives":11,"api_compatible":15,"body":20,"category":581,"chinese_friendly":570,"cover":582,"description":583,"domestic":584,"extension":585,"faq":586,"free":587,"github":562,"languages":588,"lastVerified":590,"meta":591,"models":586,"navigation":587,"notSuitable":586,"opensource":587,"path":592,"pillar":593,"platforms":594,"priceTable":586,"pricing":597,"published":598,"relatedPlaybooks":586,"relatedReviews":586,"score":599,"self_host":584,"seo":602,"seoTitle":603,"slug":604,"sources":605,"stem":608,"suitable":586,"tagline":609,"tags":610,"updated":590,"verdict":616,"website":554,"__hash__":617},"tools\u002Ftools\u002Fagent\u002Fplatform\u002Fcrewai.md","CrewAI",[12,13,14],"agent\u002Fplatform\u002Flangflow","agent\u002Fplatform\u002Fautogen","agent\u002Fplatform\u002Fn8n",[16,17,18,19],"OpenAI","Anthropic","Google","Moonshot Kimi",{"type":21,"value":22,"toc":565},"minimark",[23,28,32,35,38,97,100,154,160,164,172,195,200,223,226,268,271,407,410,458,462,494,498,504,510,516,522,525,540,543,548],[24,25,27],"h2",{"id":26},"tldr","TL;DR",[29,30,31],"p",{},"CrewAI 是开源多 Agent 协作框架（MIT 协议），用 Python 代码定义 Agent 角色（Role）、目标（Goal）、工具（Tools），组合成 Crew 执行任务序列。核心概念清晰——Agent 负责做事、Task 定义做什么、Crew 编排怎么协作。支持顺序 \u002F 层级 \u002F 自定义流程，内置 50+ 工具集成。CrewAI Enterprise 提供云端托管 + 可视化监控。",[29,33,34],{},"适合：需要多 Agent 自动化工作流的开发者、Python 技术栈团队、快速原型验证多 Agent 方案。不适合：非技术用户（用 Dify）、需要极复杂条件分支流程（用 LangGraph）、需要 GUI 可视化编排（用 Flowise）。",[24,36,37],{"id":37},"核心能力",[39,40,41,49,55,61,67,73,79,85,91],"ul",{},[42,43,44,48],"li",{},[45,46,47],"strong",{},"角色定义","：Agent = Role + Goal + Backstory + Tools，角色人设驱动行为",[42,50,51,54],{},[45,52,53],{},"任务编排","：Task 定义具体任务 + 期望输出 + 分配 Agent",[42,56,57,60],{},[45,58,59],{},"Crew 编排","：顺序执行 \u002F 层级管理 \u002F 自定义流程三种模式",[42,62,63,66],{},[45,64,65],{},"工具集成","：内置 SerperDev \u002F Firecrawl \u002F FileRead \u002F WebScraper 等 50+ 工具",[42,68,69,72],{},[45,70,71],{},"流程控制","：支持任务间依赖、条件路由、输出传递",[42,74,75,78],{},[45,76,77],{},"记忆系统","：Short-term \u002F Long-term \u002F Entity Memory，Agent 可跨任务记忆",[42,80,81,84],{},[45,82,83],{},"多模型支持","：OpenAI \u002F Claude \u002F Gemini \u002F Ollama \u002F 任意 LiteLLM 兼容模型",[42,86,87,90],{},[45,88,89],{},"CrewAI Enterprise","：云端托管 + 可视化 Crew 监控 + 团队协作",[42,92,93,96],{},[45,94,95],{},"输出结构化","：支持 Pydantic 模型定义输出格式",[24,98,99],{"id":99},"价格",[101,102,103,118],"table",{},[104,105,106],"thead",{},[107,108,109,113,115],"tr",{},[110,111,112],"th",{},"方案",[110,114,99],{},[110,116,117],{},"核心功能",[119,120,121,133,143],"tbody",{},[107,122,123,127,130],{},[124,125,126],"td",{},"开源版",[124,128,129],{},"$0",[124,131,132],{},"完整框架，MIT 协议，本地运行",[107,134,135,137,140],{},[124,136,89],{},[124,138,139],{},"$49\u002F月起",[124,141,142],{},"云端托管 + 可视化监控 + API",[107,144,145,148,151],{},[124,146,147],{},"Enterprise+",[124,149,150],{},"联系销售",[124,152,153],{},"SSO \u002F 私有部署 \u002F 专属支持",[155,156,157],"blockquote",{},[29,158,159],{},"价格信息基于 2026-07 官网，可能调整。",[24,161,163],{"id":162},"体验与评测资料整理","体验与评测（资料整理）",[155,165,166],{},[29,167,168,169],{},"说明：本节基于官方文档与公开评测整理，非本站独立实测环境，具体数据请以官方为准。\n",[45,170,171],{},"亮点：",[39,173,174,177,180,183,186,189,192],{},[42,175,176],{},"API 设计非常直观，Agent + Task + Crew 三件套 10 分钟上手",[42,178,179],{},"角色人设（Backstory）确实影响 Agent 行为，写好 backstory 效果提升明显",[42,181,182],{},"层级模式（hierarchical）下 Manager Agent 自动分配任务，适合复杂场景",[42,184,185],{},"内置工具丰富，SerperDev 搜索 + Firecrawl 爬虫开箱即用",[42,187,188],{},"Memory 系统让 Agent 跨任务保持上下文，长流程不丢信息",[42,190,191],{},"Pydantic 结构化输出对后续处理非常友好",[42,193,194],{},"CrewAI Enterprise 的可视化监控能看到每个 Agent 的思考过程",[29,196,197],{},[45,198,199],{},"踩坑：",[39,201,202,205,208,211,214,217,220],{},[42,203,204],{},"Agent 偶尔\"不听话\"——偏离角色设定、重复执行、跳过任务",[42,206,207],{},"复杂流程的调试困难，错误信息不够清晰",[42,209,210],{},"Token 消耗不小——多 Agent + 多轮对话 + Memory 存储",[42,212,213],{},"开源版无 GUI，全靠日志调试，Enterprise 版才有可视化",[42,215,216],{},"版本迭代快，API 偶有 breaking changes",[42,218,219],{},"层级模式的 Manager Agent 判断不稳定，有时分配不合理",[42,221,222],{},"中文 system prompt 效果不如英文，建议角色定义用英文",[24,224,225],{"id":225},"上手",[227,228,229,236,243,249,255,261],"ol",{},[42,230,231,235],{},[232,233,234],"code",{},"pip install crewai crewai-tools","（建议用 uv 管理虚拟环境）",[42,237,238,239,242],{},"配置 LLM：设置 ",[232,240,241],{},"OPENAI_API_KEY"," 或在代码中指定 model",[42,244,245,246],{},"定义 Agent：",[232,247,248],{},"Agent(role=\"研究员\", goal=\"搜集信息\", tools=[search_tool])",[42,250,251,252],{},"定义 Task：",[232,253,254],{},"Task(description=\"调研XX趋势\", agent=researcher, expected_output=\"报告\")",[42,256,257,258],{},"组建 Crew：",[232,259,260],{},"Crew(agents=[researcher, writer], tasks=[task1, task2], process=Process.sequential)",[42,262,263,264,267],{},"启动：",[232,265,266],{},"result = crew.kickoff()","，查看结果 + 日志",[24,269,270],{"id":270},"对比",[101,272,273,291],{},[104,274,275],{},[107,276,277,280,282,285,288],{},[110,278,279],{},"维度",[110,281,10],{},[110,283,284],{},"AutoGen",[110,286,287],{},"LangGraph",[110,289,290],{},"Dify",[119,292,293,309,326,343,359,375,390],{},[107,294,295,298,301,304,306],{},[124,296,297],{},"上手难度",[124,299,300],{},"低",[124,302,303],{},"高",[124,305,303],{},[124,307,308],{},"极低",[107,310,311,314,317,320,323],{},[124,312,313],{},"API 设计",[124,315,316],{},"优雅直观",[124,318,319],{},"底层灵活",[124,321,322],{},"图模型",[124,324,325],{},"可视化",[107,327,328,331,334,337,340],{},[124,329,330],{},"多 Agent",[124,332,333],{},"✅ Crew 角色",[124,335,336],{},"✅ Group Chat",[124,338,339],{},"✅ 图编排",[124,341,342],{},"✅ 工作流",[107,344,345,348,351,354,356],{},[124,346,347],{},"代码执行",[124,349,350],{},"需自定义",[124,352,353],{},"✅ Docker 沙箱",[124,355,350],{},[124,357,358],{},"沙箱",[107,360,361,363,366,369,372],{},[124,362,77],{},[124,364,365],{},"✅ 内置",[124,367,368],{},"有限",[124,370,371],{},"需自建",[124,373,374],{},"✅",[107,376,377,380,383,386,388],{},[124,378,379],{},"GUI",[124,381,382],{},"Enterprise 版",[124,384,385],{},"❌",[124,387,385],{},[124,389,374],{},[107,391,392,395,398,401,404],{},[124,393,394],{},"适合场景",[124,396,397],{},"业务自动化",[124,399,400],{},"研究",[124,402,403],{},"精确流程",[124,405,406],{},"应用构建",[24,408,409],{"id":409},"避坑",[39,411,412,418,424,430,436,442,452],{},[42,413,414,417],{},[45,415,416],{},"Backstory 认真写","：角色人设直接影响 Agent 行为质量，模糊描述 = 模糊行为",[42,419,420,423],{},[45,421,422],{},"expected_output 必填","：不定义预期输出，Agent 容易跑偏",[42,425,426,429],{},[45,427,428],{},"控制 Agent 数量","：3-5 个 Agent 最佳，超过 8 个协调成本急升",[42,431,432,435],{},[45,433,434],{},"Memory 按需开启","：Long-term Memory 会累积 token 消耗，简单任务关掉",[42,437,438,441],{},[45,439,440],{},"调试用 verbose=True","：开启详细日志看 Agent 思考过程，定位问题",[42,443,444,447,448,451],{},[45,445,446],{},"版本锁定","：",[232,449,450],{},"pip install crewai==x.x.x","，迭代快别用 latest",[42,453,454,457],{},[45,455,456],{},"层级模式慎用","：Manager Agent 不稳定，简单场景用 sequential 更可靠",[24,459,461],{"id":460},"适合-不适合","适合 \u002F 不适合",[39,463,464,467,470,473,476,479,482,485,488,491],{},[42,465,466],{},"✅ Python 开发者快速构建多 Agent 工作流",[42,468,469],{},"✅ 内容生产流水线（调研 → 写作 → 审核）",[42,471,472],{},"✅ 自动化研究 \u002F 数据收集 \u002F 报告生成",[42,474,475],{},"✅ 需要角色分工的协作场景",[42,477,478],{},"✅ 快速原型验证多 Agent 方案",[42,480,481],{},"❌ 非技术用户（用 Dify \u002F Flowise）",[42,483,484],{},"❌ 需要极复杂条件分支流程（用 LangGraph）",[42,486,487],{},"❌ 需要精细控制 Agent 对话轮次（用 AutoGen）",[42,489,490],{},"❌ 需要免费 GUI 可视化监控（开源版无 GUI）",[42,492,493],{},"❌ 预算敏感的高频调用场景（多 Agent token 消耗大）",[24,495,497],{"id":496},"faq","FAQ",[29,499,500,503],{},[45,501,502],{},"Q: CrewAI 和 AutoGen 怎么选？","\nA: CrewAI 上手更快，Agent + Task + Crew 概念直观，适合业务自动化和快速原型。AutoGen 更底层灵活，Group Chat + 代码执行 + 事件驱动适合研究和复杂协作。做产品选 CrewAI，做研究选 AutoGen。",[29,505,506,509],{},[45,507,508],{},"Q: 开源版和 Enterprise 版差别大吗？","\nA: 开源版框架功能完整，能跑所有 Agent \u002F Task \u002F Crew。Enterprise 版主要多了云端托管（免运维）、可视化监控（看 Agent 思考过程）、团队协作和 API 服务。如果只是本地跑 Agent 工作流，开源版够用。",[29,511,512,515],{},[45,513,514],{},"Q: 可以接入本地模型吗？","\nA: 可以。CrewAI 基于 LiteLLM，支持 Ollama \u002F vLLM \u002F LM Studio 等本地模型。但本地模型能力有限，角色扮演和工具调用效果可能不如 GPT-4 \u002F Claude。建议开发用便宜模型，生产用高级模型。",[29,517,518,521],{},[45,519,520],{},"Q: Agent 总是跑偏怎么办？","\nA: 三步排查：1）检查 Backstory 是否足够具体；2）确认 expected_output 定义清晰；3）开启 verbose=True 看思考过程定位偏移点。复杂任务拆成更小的 Task，每个 Task 目标单一明确。",[24,523,524],{"id":524},"相关阅读",[29,526,527,531,532,531,536],{},[528,529,284],"a",{"href":530},"\u002Fagent\u002Fplatform\u002Fautogen.html"," · ",[528,533,535],{"href":534},"\u002Fagent\u002Fplatform\u002Fflowise.html","Flowise",[528,537,539],{"href":538},"\u002Fcoding\u002Fapi\u002Flangfuse.html","Langfuse",[24,541,542],{"id":542},"来源",[155,544,545],{},[29,546,547],{},"本文的价格、版本号与性能数据均参考以下官方渠道整理，可能随时间变动，请以官方实时信息为准。",[39,549,550,558],{},[42,551,552],{},[528,553,557],{"href":554,"rel":555},"https:\u002F\u002Fcrewai.com",[556],"nofollow","官网",[42,559,560],{},[528,561,564],{"href":562,"rel":563},"https:\u002F\u002Fgithub.com\u002FcrewAIInc\u002FcrewAI",[556],"GitHub",{"title":566,"searchDepth":567,"depth":567,"links":568},"",3,[569,571,572,573,574,575,576,577,578,579,580],{"id":26,"depth":570,"text":27},2,{"id":37,"depth":570,"text":37},{"id":99,"depth":570,"text":99},{"id":162,"depth":570,"text":163},{"id":225,"depth":570,"text":225},{"id":270,"depth":570,"text":270},{"id":409,"depth":570,"text":409},{"id":460,"depth":570,"text":461},{"id":496,"depth":570,"text":497},{"id":524,"depth":570,"text":524},{"id":542,"depth":570,"text":542},"platform","\u002Fimg\u002Ftools\u002Fcrewai.webp","CrewAI 真实评测：开源多 Agent 协作框架（MIT 协议），用 Python 代码定义 Agent 角色、任务和协作流程。支持角色分工、任务编排、工具集成，适合需要构建多 Agent 自动化工作流的开发者和企业团队。",false,"md",null,true,[589],"en","2026-07-30",{},"\u002Ftools\u002Fagent\u002Fplatform\u002Fcrewai","agent",[595,596],"linux","docker","Free \u002F 开源（MIT）\u002F Enterprise","2026-07-05",{"power":600,"ux":567,"price":601,"cn_support":570,"stability":567},4,5,{"title":10,"description":583},"CrewAI - 多 Agent 协作框架评测与使用 | AIHO","agent\u002Fplatform\u002Fcrewai",[606,607],{"title":557,"url":554},{"title":564,"url":562},"tools\u002Fagent\u002Fplatform\u002Fcrewai","多 Agent 协作框架，Python 代码定义角色和任务",[611,612,613,614,615],"agent-platform","multi-agent","framework","python","opensource","需要用 Python 快速构建多 Agent 自动化工作流的开发者首选，角色 + 任务 + Crew 的概念直观易学、API 设计优雅，但 Agent 对话可控性和稳定性不如 AutoGen，复杂流程编排需配合 LangGraph。","oi_8bSDkK_n2D-4mhKaXx1OMPYNIsP0E3tESVED9rwo",[619,1275,1750],{"id":620,"title":621,"alternatives":622,"api_compatible":625,"body":629,"category":581,"chinese_friendly":567,"cover":1209,"description":1210,"domestic":584,"extension":585,"faq":1211,"free":587,"github":1224,"languages":1225,"lastVerified":1227,"meta":1228,"models":586,"navigation":587,"notSuitable":586,"opensource":587,"path":1229,"pillar":593,"platforms":1230,"priceTable":1234,"pricing":1248,"published":1249,"relatedPlaybooks":1250,"relatedReviews":586,"score":1252,"self_host":587,"seo":1253,"seoTitle":1254,"slug":12,"sources":1255,"stem":1266,"suitable":586,"tagline":1267,"tags":1268,"updated":1259,"verdict":1273,"website":1258,"__hash__":1274},"tools\u002Ftools\u002Fagent\u002Fplatform\u002Flangflow.md","Langflow",[14,623,624],"agent\u002Fprotocol\u002Fcomposio","agent\u002Fgeneral\u002Fopenmanus",[16,17,626,627,19,628],"百度文心","阿里通义","DeepSeek",{"type":21,"value":630,"toc":1197},[631,633,636,639,641,715,717,743,748,752,756,782,786,812,814,882,885,908,910,1053,1055,1111,1113,1139,1141,1161,1163,1193],[24,632,27],{"id":26},[29,634,635],{},"Langflow 是 2023 开源、2024 被 DataStax 收购的可视化 LangChain 画布。20,000+ GitHub stars、MIT 协议、Python 实现、与 Astra DB 向量库深度集成。差异点：拖拽式画布把 LangChain primitive 映射成节点 + RAG pipeline 原生组件 + 多 agent 工作流 + 节点可下钻到 Python 代码 + 自托管 \u002F Docker \u002F DataStax Astra 云托管 + Pinecone \u002F pgvector \u002F 主流向量库适配 + Visual GUI for building LangChain pipelines。Self-host 免费 \u002F Cloud Free + Paid ~$25\u002F月起。",[29,637,638],{},"适合：工程团队要可视化建 LangChain 流 + 不被 SaaS 锁住；合规 \u002F 数据驻留要求自托管；要 Astra DB 一站式 RAG；想用 LangChain 但讨厌纯代码协作。不适合：业务侧 + 非工程师（用 Dify \u002F Coze）；纯无代码偏好（Langflow 仍要懂 LangChain 概念）；轻量场景 + 不需要画布（直接写 LangChain 代码）。",[24,640,37],{"id":37},[39,642,643,649,655,661,667,673,679,685,691,697,703,709],{},[42,644,645,648],{},[45,646,647],{},"可视化 LangChain 画布","：节点 = primitive，连线 = 数据流",[42,650,651,654],{},[45,652,653],{},"RAG pipeline 原生组件","：Chunking \u002F Embedding \u002F VectorStore \u002F Retriever",[42,656,657,660],{},[45,658,659],{},"多 agent 工作流","：编排多 agent 协作",[42,662,663,666],{},[45,664,665],{},"Python 下钻","：任意节点可写 custom Python",[42,668,669,672],{},[45,670,671],{},"向量库适配","：Astra DB \u002F Pinecone \u002F pgvector \u002F Weaviate \u002F Chroma",[42,674,675,678],{},[45,676,677],{},"API 部署","：流程一键导出为 REST API",[42,680,681,684],{},[45,682,683],{},"Real-time collaboration","：多用户同 project",[42,686,687,690],{},[45,688,689],{},"版本控制","：内置 versioning + revert",[42,692,693,696],{},[45,694,695],{},"数据可视化","：node output \u002F data flow 可视化调试",[42,698,699,702],{},[45,700,701],{},"角色权限","：user auth + RBAC",[42,704,705,708],{},[45,706,707],{},"Docker \u002F pip 安装","：5 分钟启动",[42,710,711,714],{},[45,712,713],{},"Astra-hosted cloud","：DataStax 托管选项",[24,716,99],{"id":99},[39,718,719,725,731,737],{},[42,720,721,724],{},[45,722,723],{},"Self-Host (OSS)","：$0；MIT 完全免费，自付 LLM API + 服务器",[42,726,727,730],{},[45,728,729],{},"Cloud Free","：$0；DataStax Astra-hosted 小流量",[42,732,733,736],{},[45,734,735],{},"Cloud Paid","：~$25\u002F月起；Astra DB + 更高额度 + 团队协作",[42,738,739,742],{},[45,740,741],{},"Enterprise","：联系销售；SSO + audit + 私有部署 + SLA",[155,744,745],{},[29,746,747],{},"自托管最低成本：$5-10\u002F月 VPS + LLM API token。Cloud Paid $25\u002F月适合不想运维的小团队。",[24,749,751],{"id":750},"实测rag-应用-内部工具搭建","实测（RAG 应用 + 内部工具搭建）",[29,753,754],{},[45,755,171],{},[39,757,758,761,764,767,770,773,776,779],{},[42,759,760],{},"画布直观，比纯写 LangChain 协作效率高 5x",[42,762,763],{},"节点下钻到 Python 让灵活度不被画布限制",[42,765,766],{},"Astra DB 集成省了配 vector store 时间",[42,768,769],{},"4.4\u002F5 用户评分（Propicked \u002F Tooliverse）",[42,771,772],{},"开源 + 自托管 + 数据驻留满足合规",[42,774,775],{},"多 agent 编排比裸 LangChain 调试容易",[42,777,778],{},"RAG pipeline 模板一键起 demo",[42,780,781],{},"与 DataStax 长期支持降低 abandon ware 风险",[29,783,784],{},[45,785,199],{},[39,787,788,791,794,797,800,803,806,809],{},[42,789,790],{},"稳定性偶发：复杂流大流量下 node 偶尔失联（SelectHub 反馈）",[42,792,793],{},"第三方 API 依赖：external API 失败时错误处理弱",[42,795,796],{},"production readiness 不算 mission-critical（要自加 observability）",[42,798,799],{},"LangChain 升级偶尔 break 旧 flow",[42,801,802],{},"文档对新组件滞后 1-2 月",[42,804,805],{},"中文 UI 不完整，业务侧用户上手陡",[42,807,808],{},"大型 flow（100+ 节点）画布卡顿",[42,810,811],{},"多人协作偶发同步冲突",[24,813,225],{"id":225},[815,816,820],"pre",{"className":817,"code":818,"language":819,"meta":566,"style":566},"language-bash shiki shiki-themes github-light github-dark","# pip 安装\npip install langflow\nlangflow run  # http:\u002F\u002Flocalhost:7860\n\n# 或 Docker\ndocker run -p 7860:7860 langflowai\u002Flangflow:latest\n","bash",[232,821,822,831,844,855,860,865],{"__ignoreMap":566},[823,824,827],"span",{"class":825,"line":826},"line",1,[823,828,830],{"class":829},"sJ8bj","# pip 安装\n",[823,832,833,837,841],{"class":825,"line":570},[823,834,836],{"class":835},"sScJk","pip",[823,838,840],{"class":839},"sZZnC"," install",[823,842,843],{"class":839}," langflow\n",[823,845,846,849,852],{"class":825,"line":567},[823,847,848],{"class":835},"langflow",[823,850,851],{"class":839}," run",[823,853,854],{"class":829},"  # http:\u002F\u002Flocalhost:7860\n",[823,856,857],{"class":825,"line":600},[823,858,859],{"emptyLinePlaceholder":587},"\n",[823,861,862],{"class":825,"line":601},[823,863,864],{"class":829},"# 或 Docker\n",[823,866,868,870,872,876,879],{"class":825,"line":867},6,[823,869,596],{"class":835},[823,871,851],{"class":839},[823,873,875],{"class":874},"sj4cs"," -p",[823,877,878],{"class":839}," 7860:7860",[823,880,881],{"class":839}," langflowai\u002Flangflow:latest\n",[29,883,884],{},"试 RAG 流：",[227,886,887,890,893,896,899,902,905],{},[42,888,889],{},"新建 flow → 选 Document QA 模板",[42,891,892],{},"Document Loader 节点 → 上传 PDF",[42,894,895],{},"Splitter → Embedder（OpenAI 或本地）",[42,897,898],{},"VectorStore（Astra \u002F Chroma）",[42,900,901],{},"Retriever + ChatOpenAI → Chat Output",[42,903,904],{},"部署为 API → 拿到 endpoint",[42,906,907],{},"复杂场景下钻节点写 Python 自定义",[24,909,270],{"id":270},[101,911,912,927],{},[104,913,914],{},[107,915,916,918,920,922,925],{},[110,917,279],{},[110,919,621],{},[110,921,290],{},[110,923,924],{},"n8n",[110,926,535],{},[119,928,929,946,962,977,990,1006,1019,1036],{},[107,930,931,934,937,940,943],{},[124,932,933],{},"中心",[124,935,936],{},"LangChain primitive",[124,938,939],{},"LLMOps 全平台",[124,941,942],{},"通用 workflow",[124,944,945],{},"LangChain（JS）",[107,947,948,951,954,957,960],{},[124,949,950],{},"开源",[124,952,953],{},"✅ MIT",[124,955,956],{},"✅ AGPL",[124,958,959],{},"✅ Sustainable",[124,961,953],{},[107,963,964,967,970,973,975],{},[124,965,966],{},"自托管",[124,968,969],{},"✅ pip\u002FDocker",[124,971,972],{},"✅ Docker",[124,974,972],{},[124,976,374],{},[107,978,979,981,984,986,988],{},[124,980,325],{},[124,982,983],{},"✅ 旗舰",[124,985,374],{},[124,987,374],{},[124,989,374],{},[107,991,992,995,998,1001,1004],{},[124,993,994],{},"代码下钻",[124,996,997],{},"✅ Python",[124,999,1000],{},"部分",[124,1002,1003],{},"✅ JS",[124,1005,1003],{},[107,1007,1008,1011,1013,1015,1017],{},[124,1009,1010],{},"RAG 内置",[124,1012,374],{},[124,1014,374],{},[124,1016,1000],{},[124,1018,374],{},[107,1020,1021,1024,1027,1030,1033],{},[124,1022,1023],{},"起价（云）",[124,1025,1026],{},"$25\u002F月",[124,1028,1029],{},"$59\u002F月（Team）",[124,1031,1032],{},"自托管 $0",[124,1034,1035],{},"–",[107,1037,1038,1041,1044,1047,1050],{},[124,1039,1040],{},"适合",[124,1042,1043],{},"工程 + LangChain",[124,1045,1046],{},"业务 + LLMOps",[124,1048,1049],{},"通用自动化",[124,1051,1052],{},"JS 生态",[24,1054,409],{"id":409},[39,1056,1057,1063,1069,1075,1081,1087,1093,1099,1105],{},[42,1058,1059,1062],{},[45,1060,1061],{},"自托管推荐 Docker","：pip 版本依赖冲突难调",[42,1064,1065,1068],{},[45,1066,1067],{},"生产加 observability","：langsmith \u002F sentry \u002F prometheus 必装",[42,1070,1071,1074],{},[45,1072,1073],{},"Astra DB cloud free","：起步够用，付费版起步前算清成本",[42,1076,1077,1080],{},[45,1078,1079],{},"复杂 flow 拆模块","：100+ 节点画布卡顿，拆成子 flow",[42,1082,1083,1086],{},[45,1084,1085],{},"LangChain 版本 pin","：Langflow 升级前测试 flow 兼容性",[42,1088,1089,1092],{},[45,1090,1091],{},"第三方 API 加重试","：custom Python 节点写 retry + fallback",[42,1094,1095,1098],{},[45,1096,1097],{},"多人协作 lock","：同时编辑 flow 容易冲突，加 lock 流程",[42,1100,1101,1104],{},[45,1102,1103],{},"中文场景","：UI 英文为主，业务侧用户先培训",[42,1106,1107,1110],{},[45,1108,1109],{},"Self-host vs Cloud","：合规要 self-host，省心要 Cloud",[24,1112,461],{"id":460},[39,1114,1115,1118,1121,1124,1127,1130,1133,1136],{},[42,1116,1117],{},"✅ 工程团队要可视化建 LangChain 流",[42,1119,1120],{},"✅ 合规 \u002F 数据驻留要求自托管",[42,1122,1123],{},"✅ 要 Astra DB 一站式 RAG",[42,1125,1126],{},"✅ Python 团队 + 想画布 + 想下钻代码",[42,1128,1129],{},"❌ 业务侧 + 非工程师（用 Dify \u002F Coze）",[42,1131,1132],{},"❌ 纯无代码偏好",[42,1134,1135],{},"❌ 轻量场景 + 直接写 LangChain 更快",[42,1137,1138],{},"❌ JS 生态优先（用 Flowise）",[24,1140,524],{"id":524},[39,1142,1143,1149,1155],{},[42,1144,1145],{},[528,1146,1148],{"href":1147},"\u002Ftools\u002Fagent\u002Fplatform\u002Fn8n","n8n 评测",[42,1150,1151],{},[528,1152,1154],{"href":1153},"\u002Ftools\u002Fagent\u002Fprotocol\u002Fcomposio","Composio 评测",[42,1156,1157],{},[528,1158,1160],{"href":1159},"\u002Ftools\u002Fagent\u002Fgeneral\u002Fopenmanus","OpenManus 评测",[24,1162,542],{"id":542},[227,1164,1165,1172,1179,1186],{},[42,1166,1167,1168],{},"Langflow 官网 + 定价 ",[528,1169,1170],{"href":1170,"rel":1171},"https:\u002F\u002Fwww.langflow.org\u002Fpricing",[556],[42,1173,1174,1175],{},"AutomationAtlas — Langflow $25\u002Fmo 2026 Comparison ",[528,1176,1177],{"href":1177,"rel":1178},"https:\u002F\u002Fautomationatlas.io\u002Ftools\u002Flangflow",[556],[42,1180,1181,1182],{},"xpay — LangFlow 2026 Visual AI Agent Builder（MIT + DataStax）",[528,1183,1184],{"href":1184,"rel":1185},"https:\u002F\u002Fwww.xpay.sh\u002Fresources\u002Fagentic-frameworks\u002Flangflow",[556],[42,1187,1188,1189],{},"SelectHub — LangFlow Reviews 2026（稳定性 + 生产 readiness）",[528,1190,1191],{"href":1191,"rel":1192},"https:\u002F\u002Fwww.selecthub.com\u002Fp\u002Fai-agent-builder-software\u002Flangflow\u002F",[556],[1194,1195,1196],"style",{},"html pre.shiki code .sJ8bj, html code.shiki .sJ8bj{--shiki-default:#6A737D;--shiki-dark:#6A737D}html pre.shiki code .sScJk, html code.shiki .sScJk{--shiki-default:#6F42C1;--shiki-dark:#B392F0}html pre.shiki code .sZZnC, html code.shiki .sZZnC{--shiki-default:#032F62;--shiki-dark:#9ECBFF}html pre.shiki code .sj4cs, html code.shiki .sj4cs{--shiki-default:#005CC5;--shiki-dark:#79B8FF}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html.dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}",{"title":566,"searchDepth":567,"depth":567,"links":1198},[1199,1200,1201,1202,1203,1204,1205,1206,1207,1208],{"id":26,"depth":570,"text":27},{"id":37,"depth":570,"text":37},{"id":99,"depth":570,"text":99},{"id":750,"depth":570,"text":751},{"id":225,"depth":570,"text":225},{"id":270,"depth":570,"text":270},{"id":409,"depth":570,"text":409},{"id":460,"depth":570,"text":461},{"id":524,"depth":570,"text":524},{"id":542,"depth":570,"text":542},"\u002Fimg\u002Ftools\u002Flangflow.webp","Langflow 真实评测：2023 开源项目，2024 被 DataStax 收购，与 Astra DB 向量数据库平台深度集成。20,000+ GitHub stars、MIT 协议。差异点：拖拽式可视化画布映射 LangChain 原语 + RAG pipeline 原生组件 + 多 agent 工作流 + Python + 自托管（pip \u002F Docker）+ DataStax Astra 托管云 + Pinecone \u002F pgvector \u002F 主流向量库适配。Cloud 免费档 + 约 $25\u002F月起。",[1212,1215,1218,1221],{"q":1213,"a":1214},"Langflow 和 Dify \u002F n8n \u002F Flowise 怎么选？","Langflow 强『可视化 LangChain 原语 + Python 代码可下钻 + 自托管 + Astra DB 集成』，工程导向。Dify 是 LLMOps 全平台（含 dataset \u002F app \u002F observability），业务侧更友好。n8n 是通用 workflow（非 LangChain 中心），70+ LangChain 节点是后加的。Flowise 是 Langflow 的同类竞品（JS 生态）。工程团队 + Python + 合规自托管 → Langflow；业务 + 完整 LLMOps → Dify；通用自动化 → n8n。",{"q":1216,"a":1217},"为什么被 DataStax 收购？","DataStax 是 Cassandra 商业公司 + Astra DB 向量数据库厂商。收购 Langflow 是为了把『可视化 LangChain builder』和『生产级 vector store』捆成一站式 RAG 解决方案。Langflow 主仓仍是 MIT 开源，但 cloud \u002F 企业版深度依赖 Astra。",{"q":1219,"a":1220},"可视化画布会不会限制灵活度？","Langflow 节点对应 LangChain primitive，可在任意节点下钻到 Python 代码 + 自定义。所以可视化层是『脚手架 + 协作工具』，不是『纯无代码黑盒』。复杂逻辑、custom tool、retriever 都能写代码扩展。",{"q":1222,"a":1223},"production readiness 如何？","用户反馈：原型 + 内部工具非常顺；大流量 \u002F 关键业务要自行加 observability \u002F 错误处理 \u002F 缓存。SelectHub 评测列出『稳定性偶发 + 第三方 API 依赖 + 非完全 production-ready』。生产部署建议 Astra-hosted cloud 或自托管 + 加 sentry \u002F langsmith \u002F prometheus。","https:\u002F\u002Fgithub.com\u002Flangflow-ai\u002Flangflow",[589,1226],"multi","2026-08-02",{},"\u002Ftools\u002Fagent\u002Fplatform\u002Flangflow",[1231,1232,596,1233],"self-host","cloud","web",[1235,1238,1241,1245],{"plan":723,"price":129,"features":1236,"notes":1237},"MIT + pip\u002FDocker + 全部组件 + Pinecone\u002Fpgvector\u002FAstra 适配","自付 LLM API + 服务器",{"plan":729,"price":129,"features":1239,"notes":1240},"DataStax 托管 + 小流量","试水",{"plan":735,"price":1242,"features":1243,"notes":1244},"~$25\u002F月起","Astra DB + 更高额度 + 更多并发 + 团队协作","按使用量阶梯",{"plan":741,"price":150,"features":1246,"notes":1247},"SSO + audit + 私有部署 + SLA + 数据驻留","合规 \u002F 大客户","Self-host 完全免费 MIT \u002F Cloud Free + 付费起 ~$25·月（Astra-hosted）","2026-06-19",[1251],"onboarding\u002Frag-app-workflow",{"power":600,"ux":601,"price":601,"cn_support":567,"stability":600},{"title":621,"description":1210},"Langflow 评测 2026：可视化 AI 工作流构建工具，LangChain 低代码平台",[1256,1260,1262,1264],{"name":1257,"url":1258,"accessed":1259},"Langflow 官网","https:\u002F\u002Fwww.langflow.org","2026-06-24",{"name":1261,"url":1177,"accessed":1259},"AutomationAtlas — Langflow $25\u002Fmo 2026 Comparison",{"name":1263,"url":1184,"accessed":1259},"xpay — LangFlow 2026 Visual AI Agent Builder",{"name":1265,"url":1191,"accessed":1259},"SelectHub — LangFlow Reviews 2026","tools\u002Fagent\u002Fplatform\u002Flangflow","DataStax 出品的可视化 LangChain 画布——MIT 开源 + 20k+ stars + 自托管 + Astra DB 云托管",[615,1269,1270,1271,1272,848],"visual-builder","langchain","rag","datastax","工程团队要可视化建 LangChain 流 + 不被 SaaS 锁住 + 合规要求自托管的最佳选择。要纯无代码 + 业务侧 → Dify \u002F Coze；要纯代码 → 直接写 LangChain \u002F LlamaIndex。","PNaJCu7eJDsH8LhAlO7CJ5JmktlB6hCq0vbBojxlTeM",{"id":1276,"title":284,"alternatives":1277,"api_compatible":1279,"body":1280,"category":581,"chinese_friendly":570,"cover":1731,"description":1732,"domestic":584,"extension":585,"faq":586,"free":587,"github":1716,"languages":1733,"lastVerified":590,"meta":1734,"models":586,"navigation":587,"notSuitable":586,"opensource":587,"path":1735,"pillar":593,"platforms":1736,"priceTable":586,"pricing":1737,"published":598,"relatedPlaybooks":586,"relatedReviews":586,"score":1738,"self_host":584,"seo":1739,"seoTitle":1740,"slug":13,"sources":1741,"stem":1744,"suitable":586,"tagline":1745,"tags":1746,"updated":590,"verdict":1748,"website":1710,"__hash__":1749},"tools\u002Ftools\u002Fagent\u002Fplatform\u002Fautogen.md",[604,12,1278],"agent\u002Fplatform\u002Fdify",[16,17,18],{"type":21,"value":1281,"toc":1718},[1282,1284,1291,1294,1296,1350,1352,1355,1357,1363,1387,1391,1414,1416,1449,1451,1564,1566,1614,1616,1647,1649,1655,1673,1679,1687,1689,1698,1700,1704],[24,1283,27],{"id":26},[29,1285,1286,1287,1290],{},"AutoGen 是微软研究院开源的多 Agent 对话框架（MIT 协议），用 Python 代码定义 Agent 角色、对话流程和工具调用。核心是 ",[232,1288,1289],{},"ConversableAgent"," + Group Chat 模式——多个 Agent 自动对话协作完成任务，支持代码执行、工具调用、人在回路（human-in-the-loop）。AutoGen 0.4+ 重构为事件驱动架构，性能和扩展性大幅提升。",[29,1292,1293],{},"适合：AI 研究者、需要精细控制 Agent 协作逻辑的高级开发者、多 Agent 实验项目。不适合：快速原型验证（用 CrewAI）、非技术用户（用 Dify \u002F Flowise）、需要 GUI 的场景、追求 API 稳定性的生产项目。",[24,1295,37],{"id":37},[39,1297,1298,1304,1309,1315,1321,1327,1333,1338,1344],{},[42,1299,1300,1303],{},[45,1301,1302],{},"多 Agent 对话","：Group Chat 模式，多个 Agent 自动对话协作完成任务",[42,1305,1306,1308],{},[45,1307,347],{},"：内置 Docker 代码执行器，Agent 可写代码 + 运行 + 调试",[42,1310,1311,1314],{},[45,1312,1313],{},"工具调用","：自定义函数工具，Agent 自动选择和调用",[42,1316,1317,1320],{},[45,1318,1319],{},"人在回路","：Human-in-the-loop 模式，关键决策需人工确认",[42,1322,1323,1326],{},[45,1324,1325],{},"事件驱动架构","：0.4+ 重构为 async 事件驱动，支持分布式 Agent",[42,1328,1329,1332],{},[45,1330,1331],{},"可定制 Agent","：system message \u002F 工具集 \u002F 终止条件全可自定义",[42,1334,1335,1337],{},[45,1336,83],{},"：OpenAI \u002F Claude \u002F Azure \u002F Ollama \u002F Gemini 等",[42,1339,1340,1343],{},[45,1341,1342],{},"Agent 可组合","：嵌套 Agent、层级 Agent、条件路由",[42,1345,1346,1349],{},[45,1347,1348],{},"可观测性","：集成 OpenTelemetry \u002F LangSmith 追踪 Agent 行为",[24,1351,99],{"id":99},[29,1353,1354],{},"完全免费、MIT 开源、商用免费。运行成本仅来自所接入的 LLM API 调用费用。",[24,1356,163],{"id":162},[155,1358,1359],{},[29,1360,168,1361],{},[45,1362,171],{},[39,1364,1365,1368,1371,1374,1381,1384],{},[42,1366,1367],{},"Group Chat 模式让多 Agent 协作真正\"自动化\"，代码 reviewer + coder + tester 角色分工清晰",[42,1369,1370],{},"Docker 代码执行器安全隔离，Agent 写的代码在沙箱中运行",[42,1372,1373],{},"0.4+ 的事件驱动架构性能提升明显，异步并发能力强",[42,1375,1376,1377,1380],{},"自定义工具集成灵活，Python 函数加 ",[232,1378,1379],{},"@user_function"," 装饰器即可",[42,1382,1383],{},"微软背书，学术认可度高，论文引用多",[42,1385,1386],{},"人在回路模式适合需要人工把关的高风险场景",[29,1388,1389],{},[45,1390,199],{},[39,1392,1393,1396,1399,1402,1405,1408,1411],{},[42,1394,1395],{},"学习曲线非常陡峭，文档虽全但概念密集，新手容易劝退",[42,1397,1398],{},"0.2 → 0.4 API 大改，迁移成本高，网上旧教程大量失效",[42,1400,1401],{},"Agent 对话容易\"跑飞\"——无限循环 \u002F 偏离主题，需仔细设计终止条件",[42,1403,1404],{},"无 GUI，调试全靠日志和 print，排查多 Agent 对话链路费时",[42,1406,1407],{},"Token 消耗大——多 Agent 对话轮次多，API 费用叠加明显",[42,1409,1410],{},"错误处理不够健壮，LLM 返回格式异常时容易崩溃",[42,1412,1413],{},"社区活跃度不如 LangChain \u002F CrewAI，遇到问题搜索不到答案",[24,1415,225],{"id":225},[227,1417,1418,1424,1429,1434,1440,1446],{},[42,1419,1420,1423],{},[232,1421,1422],{},"pip install autogen-agentchat autogen-ext","（0.4+ 新包名）",[42,1425,238,1426,1428],{},[232,1427,241],{}," 环境变量或代码内传入",[42,1430,245,1431],{},[232,1432,1433],{},"AssistantAgent(name=\"coder\", system_message=\"...\", model_client=...)",[42,1435,1436,1437],{},"创建 Group Chat：",[232,1438,1439],{},"RoundRobinGroupChat(agents=[agent1, agent2])",[42,1441,1442,1443],{},"发起任务：",[232,1444,1445],{},"result = await team.run(task=\"写一个贪吃蛇游戏\")",[42,1447,1448],{},"进阶：加 Docker 代码执行器 + 自定义工具 + 人在回路",[24,1450,270],{"id":270},[101,1452,1453,1467],{},[104,1454,1455],{},[107,1456,1457,1459,1461,1463,1465],{},[110,1458,279],{},[110,1460,284],{},[110,1462,10],{},[110,1464,287],{},[110,1466,290],{},[119,1468,1469,1484,1497,1509,1521,1534,1550],{},[107,1470,1471,1474,1477,1479,1481],{},[124,1472,1473],{},"形态",[124,1475,1476],{},"Python 框架",[124,1478,1476],{},[124,1480,1476],{},[124,1482,1483],{},"可视化平台",[107,1485,1486,1488,1490,1493,1495],{},[124,1487,297],{},[124,1489,303],{},[124,1491,1492],{},"中",[124,1494,303],{},[124,1496,300],{},[107,1498,1499,1501,1503,1505,1507],{},[124,1500,330],{},[124,1502,336],{},[124,1504,333],{},[124,1506,339],{},[124,1508,342],{},[107,1510,1511,1513,1515,1517,1519],{},[124,1512,347],{},[124,1514,353],{},[124,1516,350],{},[124,1518,350],{},[124,1520,358],{},[107,1522,1523,1525,1527,1530,1532],{},[124,1524,379],{},[124,1526,385],{},[124,1528,1529],{},"❌（有 CrewAI Studio）",[124,1531,385],{},[124,1533,374],{},[107,1535,1536,1539,1542,1545,1548],{},[124,1537,1538],{},"API 稳定性",[124,1540,1541],{},"一般（大改过）",[124,1543,1544],{},"较好",[124,1546,1547],{},"好",[124,1549,1547],{},[107,1551,1552,1554,1557,1559,1562],{},[124,1553,394],{},[124,1555,1556],{},"研究 \u002F 复杂协作",[124,1558,397],{},[124,1560,1561],{},"精确流程控制",[124,1563,406],{},[24,1565,409],{"id":409},[39,1567,1568,1578,1584,1590,1596,1602,1608],{},[42,1569,1570,1573,1574,1577],{},[45,1571,1572],{},"锁定版本","：0.2 和 0.4 API 不兼容，",[232,1575,1576],{},"pip install"," 时务必指定版本",[42,1579,1580,1583],{},[45,1581,1582],{},"设计终止条件","：Group Chat 不设终止条件会无限对话，设 max_turns + 终止关键词",[42,1585,1586,1589],{},[45,1587,1588],{},"Token 成本控制","：多 Agent 对话 token 消耗是单 Agent 的 3-5 倍，用 GPT-4 级模型注意费用",[42,1591,1592,1595],{},[45,1593,1594],{},"代码执行器一定要用 Docker","：直接本地执行 Agent 生成的代码有安全风险",[42,1597,1598,1601],{},[45,1599,1600],{},"别指望第一次跑通","：system message 调试 + 工具定义 + 终止条件需要反复迭代",[42,1603,1604,1607],{},[45,1605,1606],{},"错误处理要完善","：LLM 返回异常格式时手动 catch + 重试",[42,1609,1610,1613],{},[45,1611,1612],{},"不要用旧教程","：0.4+ 完全重构，网上大部分 AutoGen 教程是 0.2 版本的",[24,1615,461],{"id":460},[39,1617,1618,1621,1624,1627,1630,1633,1636,1638,1641,1644],{},[42,1619,1620],{},"✅ AI 研究者实验多 Agent 协作模式",[42,1622,1623],{},"✅ 需要代码级精细控制 Agent 行为",[42,1625,1626],{},"✅ 需要 Agent 代码执行 + 自动调试",[42,1628,1629],{},"✅ 人在回路的高风险决策场景",[42,1631,1632],{},"✅ 学术项目 \u002F 论文复现",[42,1634,1635],{},"❌ 快速原型验证（用 CrewAI，API 更简洁）",[42,1637,481],{},[42,1639,1640],{},"❌ 追求 API 稳定性的生产项目（版本变动大）",[42,1642,1643],{},"❌ 需要可视化调试（无 GUI，全靠日志）",[42,1645,1646],{},"❌ 预算敏感场景（多 Agent 对话 token 消耗大）",[24,1648,497],{"id":496},[29,1650,1651,1654],{},[45,1652,1653],{},"Q: AutoGen 和 CrewAI 怎么选？","\nA: AutoGen 更底层、更灵活，适合研究和复杂多 Agent 协作实验，但学习成本高。CrewAI API 更简洁直观，角色 + 任务 + 流程的概念更易理解，适合业务自动化场景。研究选 AutoGen，做产品选 CrewAI。",[29,1656,1657,1660,1661,1664,1665,1668,1669,1672],{},[45,1658,1659],{},"Q: AutoGen 0.2 和 0.4 有什么区别？","\nA: 0.4 是完全重构版本——从同步改为异步事件驱动架构，包名从 ",[232,1662,1663],{},"pyautogen"," 改为 ",[232,1666,1667],{},"autogen-agentchat"," + ",[232,1670,1671],{},"autogen-ext","，API 全面更新。性能和扩展性大幅提升但旧代码无法直接迁移。新项目直接用 0.4+。",[29,1674,1675,1678],{},[45,1676,1677],{},"Q: 多 Agent 对话成本高吗？","\nA: 高。多 Agent 每轮对话都消耗 token，一个任务 5-10 轮对话是常态，使用 GPT-4 级模型单个任务可能花费 $0.5-2。建议开发调试用便宜模型（GPT-4o-mini），生产再切高级模型。",[29,1680,1681,1683,1684,1686],{},[45,1682,514],{},"\nA: 可以。通过 ",[232,1685,1671],{}," 的 OpenAI 兼容客户端接入 Ollama \u002F vLLM \u002F LM Studio 的本地模型端点。但本地模型能力有限，复杂多 Agent 协作效果可能不如 GPT-4 \u002F Claude。",[24,1688,524],{"id":524},[29,1690,1691,531,1694,531,1696],{},[528,1692,10],{"href":1693},"\u002Fagent\u002Fplatform\u002Fcrewai.html",[528,1695,535],{"href":534},[528,1697,539],{"href":538},[24,1699,542],{"id":542},[155,1701,1702],{},[29,1703,547],{},[39,1705,1706,1712],{},[42,1707,1708],{},[528,1709,557],{"href":1710,"rel":1711},"https:\u002F\u002Fmicrosoft.github.io\u002Fautogen",[556],[42,1713,1714],{},[528,1715,564],{"href":1716,"rel":1717},"https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Fautogen",[556],{"title":566,"searchDepth":567,"depth":567,"links":1719},[1720,1721,1722,1723,1724,1725,1726,1727,1728,1729,1730],{"id":26,"depth":570,"text":27},{"id":37,"depth":570,"text":37},{"id":99,"depth":570,"text":99},{"id":162,"depth":570,"text":163},{"id":225,"depth":570,"text":225},{"id":270,"depth":570,"text":270},{"id":409,"depth":570,"text":409},{"id":460,"depth":570,"text":461},{"id":496,"depth":570,"text":497},{"id":524,"depth":570,"text":524},{"id":542,"depth":570,"text":542},"\u002Fimg\u002Ftools\u002Fautogen.webp","AutoGen 真实评测：微软开源的多 Agent 对话框架（MIT 协议），通过代码定义 Agent 角色和协作流程，支持多 Agent 对话、工具调用、代码执行。适合需要精细控制多 Agent 协作逻辑的开发者和研究团队。",[589],{},"\u002Ftools\u002Fagent\u002Fplatform\u002Fautogen",[595,596],"Free \u002F 开源（MIT）",{"power":600,"ux":570,"price":601,"cn_support":570,"stability":567},{"title":284,"description":1732},"AutoGen - 微软多 Agent 框架评测与使用 | AIHO",[1742,1743],{"title":557,"url":1710},{"title":564,"url":1716},"tools\u002Fagent\u002Fplatform\u002Fautogen","微软开源多 Agent 对话框架，代码驱动 Agent 协作",[611,612,613,1747,614,615],"microsoft","需要代码级精细控制多 Agent 协作逻辑的研究者和高级开发者首选，微软背书 + 代码执行 + Group Chat 模式强大，但学习曲线陡峭、API 稳定性一般、无 GUI，不适合快速原型或非技术用户。","B_cns-NSiPcj5XmWQT-A-_3OiFT01qTSnk0-C6rpCWQ",{"id":1751,"title":924,"alternatives":1752,"api_compatible":1753,"body":1754,"category":581,"chinese_friendly":567,"cover":2415,"description":2416,"domestic":584,"extension":585,"faq":2417,"free":587,"github":2430,"languages":2431,"lastVerified":1227,"meta":2432,"models":586,"navigation":587,"notSuitable":586,"opensource":587,"path":1147,"pillar":593,"platforms":2433,"priceTable":2434,"pricing":2450,"published":1249,"relatedPlaybooks":2451,"relatedReviews":586,"score":2453,"self_host":587,"seo":2454,"seoTitle":2455,"slug":14,"sources":2456,"stem":2465,"suitable":586,"tagline":2466,"tags":2467,"updated":1259,"verdict":2471,"website":2377,"__hash__":2472},"tools\u002Ftools\u002Fagent\u002Fplatform\u002Fn8n.md",[12,623,624],[16,17,18],{"type":21,"value":1755,"toc":2403},[1756,1758,1761,1764,1766,1840,1842,1867,1872,1876,1880,1906,1910,1936,1940,2075,2078,2106,2109,2111,2260,2262,2323,2325,2351,2353,2368,2370,2400],[24,1757,27],{"id":26},[29,1759,1760],{},"n8n 是 2019 创立、Sustainable Use License（fair-code）的开源自动化平台。2026 突破 200,000 active users、5x ARR 增长、5,800+ 社区 AI workflow。差异点：近 70 个 LangChain 专属节点 + 原生 MCP 协议 + AI Agent 节点（reasoning loop + 工具调用）+ Ollama \u002F OpenAI 双路 + 400+ 集成 + execution-based 定价（步骤数无关）+ 自托管 VPS $5-10\u002F月跑全套。",[29,1762,1763],{},"适合：开发者 + 想完全控制 + 高量级自动化；从 Zapier \u002F Make 迁出降本；要 AI Agent + LangChain + MCP 一体；合规 \u002F 自托管 \u002F 数据驻留要求。不适合：非技术 + 要 8000+ 现成集成（用 Zapier）；中等复杂 + 不愿自托管（用 Make）；纯研究 \u002F 学术 agent（用 OpenManus \u002F Langflow）。",[24,1765,37],{"id":37},[39,1767,1768,1774,1780,1786,1792,1798,1804,1810,1816,1822,1828,1834],{},[42,1769,1770,1773],{},[45,1771,1772],{},"AI Agent 节点","：reasoning loop + 自主工具选择",[42,1775,1776,1779],{},[45,1777,1778],{},"70 LangChain 节点","：LLM \u002F VectorStore \u002F Agent \u002F Tool \u002F Memory 全套",[42,1781,1782,1785],{},[45,1783,1784],{},"原生 MCP","：MCP server 一键挂载到 agent",[42,1787,1788,1791],{},[45,1789,1790],{},"Ollama 集成","：本地 LLM 零 API 成本",[42,1793,1794,1797],{},[45,1795,1796],{},"400+ 集成","：Slack \u002F GitHub \u002F Google \u002F Notion \u002F 主流 SaaS",[42,1799,1800,1803],{},[45,1801,1802],{},"HTTP \u002F Webhook 万能节点","：任意 REST API 都能接",[42,1805,1806,1809],{},[45,1807,1808],{},"Cron \u002F Trigger","：定时 \u002F 事件 \u002F Webhook 触发",[42,1811,1812,1815],{},[45,1813,1814],{},"多分支并行 + 错误处理","：production workflow 必备",[42,1817,1818,1821],{},[45,1819,1820],{},"版本控制 + Git Sync","：workflow as code",[42,1823,1824,1827],{},[45,1825,1826],{},"自托管 Docker \u002F Kubernetes","：一行起 + 水平扩展",[42,1829,1830,1833],{},[45,1831,1832],{},"execution-based 定价","：20 步和 2 步同价（自托管 = 0）",[42,1835,1836,1839],{},[45,1837,1838],{},"5800+ 社区 workflow","：clone 即用",[24,1841,99],{"id":99},[39,1843,1844,1850,1856,1862],{},[42,1845,1846,1849],{},[45,1847,1848],{},"Community Self-host","：$0；全功能 + 不限 execution + 自付 VPS $5-10\u002F月",[42,1851,1852,1855],{},[45,1853,1854],{},"Cloud Starter","：~€20\u002F月；2,500 executions + 5 workflow",[42,1857,1858,1861],{},[45,1859,1860],{},"Cloud Pro","：~€50\u002F月；高 executions + 团队协作",[42,1863,1864,1866],{},[45,1865,741],{},"：联系销售；SSO + LDAP + 私有部署 + SLA",[155,1868,1869],{},[29,1870,1871],{},"真实场景：Zapier $50\u002F月跑中等复杂 → n8n 自托管 $5\u002F月跑同样的 = 10x 降本。",[24,1873,1875],{"id":1874},"实测中型团队-saas-迁移-本地-ai","实测（中型团队 SaaS 迁移 + 本地 AI）",[29,1877,1878],{},[45,1879,171],{},[39,1881,1882,1885,1888,1891,1894,1897,1900,1903],{},[42,1883,1884],{},"自托管成本几乎可忽略：$5\u002F月 VPS 跑几十个 workflow",[42,1886,1887],{},"AI Agent + Ollama 让 LLM 任务零 API 成本",[42,1889,1890],{},"70 LangChain 节点覆盖 RAG \u002F Agent \u002F 多模态",[42,1892,1893],{},"MCP 原生集成让 n8n agent 调用任意 MCP server",[42,1895,1896],{},"5800+ 社区 workflow 节省 80% 上手时间",[42,1898,1899],{},"HTTP \u002F Webhook 万能节点弥补 native 集成缺口",[42,1901,1902],{},"升级 Docker tag 一行，无 vendor 升级费",[42,1904,1905],{},"中文社区 \u002F B 站教程丰富",[29,1907,1908],{},[45,1909,199],{},[39,1911,1912,1915,1918,1921,1924,1927,1930,1933],{},[42,1913,1914],{},"集成数（400+）远少于 Zapier（8000+），冷门 SaaS 要写 HTTP 自己接",[42,1916,1917],{},"自托管要懂 Docker \u002F Postgres \u002F Redis（高吞吐场景）",[42,1919,1920],{},"Cloud 定价 execution 计算法与本地不一致，迁移要重算成本",[42,1922,1923],{},"复杂 workflow 调试比 Zapier 难（错误堆栈深）",[42,1925,1926],{},"Sustainable Use License 不是传统 OSI 开源，企业法务要看条款",[42,1928,1929],{},"AI Agent 节点 production 稳定性不如简单线性 flow",[42,1931,1932],{},"Webhook 公网暴露要加 IP 白名单 + secret",[42,1934,1935],{},"Worker 模式才能并发，单进程吞吐有限",[24,1937,1939],{"id":1938},"上手docker-5-分钟","上手（Docker 5 分钟）",[815,1941,1943],{"className":817,"code":1942,"language":819,"meta":566,"style":566},"# 持久化目录\nmkdir -p ~\u002Fn8n-data\n\n# 启动\ndocker run -d \\\n  --name n8n \\\n  -p 5678:5678 \\\n  -e N8N_BASIC_AUTH_ACTIVE=true \\\n  -e N8N_BASIC_AUTH_USER=admin \\\n  -e N8N_BASIC_AUTH_PASSWORD=yourpassword \\\n  -v ~\u002Fn8n-data:\u002Fhome\u002Fnode\u002F.n8n \\\n  --restart always \\\n  n8nio\u002Fn8n\n\n# http:\u002F\u002Flocalhost:5678\n",[232,1944,1945,1950,1960,1964,1969,1981,1991,2002,2016,2026,2036,2047,2058,2064,2069],{"__ignoreMap":566},[823,1946,1947],{"class":825,"line":826},[823,1948,1949],{"class":829},"# 持久化目录\n",[823,1951,1952,1955,1957],{"class":825,"line":570},[823,1953,1954],{"class":835},"mkdir",[823,1956,875],{"class":874},[823,1958,1959],{"class":839}," ~\u002Fn8n-data\n",[823,1961,1962],{"class":825,"line":567},[823,1963,859],{"emptyLinePlaceholder":587},[823,1965,1966],{"class":825,"line":600},[823,1967,1968],{"class":829},"# 启动\n",[823,1970,1971,1973,1975,1978],{"class":825,"line":601},[823,1972,596],{"class":835},[823,1974,851],{"class":839},[823,1976,1977],{"class":874}," -d",[823,1979,1980],{"class":874}," \\\n",[823,1982,1983,1986,1989],{"class":825,"line":867},[823,1984,1985],{"class":874},"  --name",[823,1987,1988],{"class":839}," n8n",[823,1990,1980],{"class":874},[823,1992,1994,1997,2000],{"class":825,"line":1993},7,[823,1995,1996],{"class":874},"  -p",[823,1998,1999],{"class":839}," 5678:5678",[823,2001,1980],{"class":874},[823,2003,2005,2008,2011,2014],{"class":825,"line":2004},8,[823,2006,2007],{"class":874},"  -e",[823,2009,2010],{"class":839}," N8N_BASIC_AUTH_ACTIVE=",[823,2012,2013],{"class":874},"true",[823,2015,1980],{"class":874},[823,2017,2019,2021,2024],{"class":825,"line":2018},9,[823,2020,2007],{"class":874},[823,2022,2023],{"class":839}," N8N_BASIC_AUTH_USER=admin",[823,2025,1980],{"class":874},[823,2027,2029,2031,2034],{"class":825,"line":2028},10,[823,2030,2007],{"class":874},[823,2032,2033],{"class":839}," N8N_BASIC_AUTH_PASSWORD=yourpassword",[823,2035,1980],{"class":874},[823,2037,2039,2042,2045],{"class":825,"line":2038},11,[823,2040,2041],{"class":874},"  -v",[823,2043,2044],{"class":839}," ~\u002Fn8n-data:\u002Fhome\u002Fnode\u002F.n8n",[823,2046,1980],{"class":874},[823,2048,2050,2053,2056],{"class":825,"line":2049},12,[823,2051,2052],{"class":874},"  --restart",[823,2054,2055],{"class":839}," always",[823,2057,1980],{"class":874},[823,2059,2061],{"class":825,"line":2060},13,[823,2062,2063],{"class":839},"  n8nio\u002Fn8n\n",[823,2065,2067],{"class":825,"line":2066},14,[823,2068,859],{"emptyLinePlaceholder":587},[823,2070,2072],{"class":825,"line":2071},15,[823,2073,2074],{"class":829},"# http:\u002F\u002Flocalhost:5678\n",[29,2076,2077],{},"连 Ollama：",[815,2079,2081],{"className":817,"code":2080,"language":819,"meta":566,"style":566},"ollama serve\nollama pull llama3.2\n# n8n 中添加 Ollama credential：http:\u002F\u002Fhost.docker.internal:11434\n",[232,2082,2083,2091,2101],{"__ignoreMap":566},[823,2084,2085,2088],{"class":825,"line":826},[823,2086,2087],{"class":835},"ollama",[823,2089,2090],{"class":839}," serve\n",[823,2092,2093,2095,2098],{"class":825,"line":570},[823,2094,2087],{"class":835},[823,2096,2097],{"class":839}," pull",[823,2099,2100],{"class":839}," llama3.2\n",[823,2102,2103],{"class":825,"line":567},[823,2104,2105],{"class":829},"# n8n 中添加 Ollama credential：http:\u002F\u002Fhost.docker.internal:11434\n",[29,2107,2108],{},"试 workflow：Webhook 触发 → AI Agent 节点（Ollama）→ Slack 通知。复制粘贴一个社区 workflow 30 分钟跑通完整 AI 自动化。",[24,2110,270],{"id":270},[101,2112,2113,2129],{},[104,2114,2115],{},[107,2116,2117,2119,2121,2124,2127],{},[110,2118,279],{},[110,2120,924],{},[110,2122,2123],{},"Zapier",[110,2125,2126],{},"Make",[110,2128,621],{},[119,2130,2131,2144,2156,2173,2186,2201,2214,2228,2244],{},[107,2132,2133,2135,2138,2140,2142],{},[124,2134,950],{},[124,2136,2137],{},"✅ fair-code",[124,2139,385],{},[124,2141,385],{},[124,2143,953],{},[107,2145,2146,2148,2150,2152,2154],{},[124,2147,966],{},[124,2149,983],{},[124,2151,385],{},[124,2153,385],{},[124,2155,374],{},[107,2157,2158,2161,2164,2167,2170],{},[124,2159,2160],{},"集成数",[124,2162,2163],{},"400+",[124,2165,2166],{},"8000+",[124,2168,2169],{},"2000+",[124,2171,2172],{},"LangChain 原语",[107,2174,2175,2178,2180,2182,2184],{},[124,2176,2177],{},"AI Agent",[124,2179,983],{},[124,2181,1000],{},[124,2183,1000],{},[124,2185,374],{},[107,2187,2188,2191,2194,2196,2198],{},[124,2189,2190],{},"LangChain 节点",[124,2192,2193],{},"✅ 70 个",[124,2195,385],{},[124,2197,385],{},[124,2199,2200],{},"✅ 原生",[107,2202,2203,2206,2208,2210,2212],{},[124,2204,2205],{},"MCP",[124,2207,2200],{},[124,2209,385],{},[124,2211,385],{},[124,2213,1000],{},[107,2215,2216,2219,2222,2224,2226],{},[124,2217,2218],{},"Local LLM",[124,2220,2221],{},"✅ Ollama",[124,2223,385],{},[124,2225,385],{},[124,2227,374],{},[107,2229,2230,2233,2236,2239,2242],{},[124,2231,2232],{},"起价",[124,2234,2235],{},"$0 自托管",[124,2237,2238],{},"$29.99\u002F月",[124,2240,2241],{},"$9\u002F月",[124,2243,2235],{},[107,2245,2246,2248,2251,2254,2257],{},[124,2247,1040],{},[124,2249,2250],{},"开发者 + 高量级",[124,2252,2253],{},"非技术 + 简单",[124,2255,2256],{},"中等复杂",[124,2258,2259],{},"LangChain 工程",[24,2261,409],{"id":409},[39,2263,2264,2270,2276,2282,2288,2294,2300,2306,2312,2317],{},[42,2265,2266,2269],{},[45,2267,2268],{},"自托管装 Postgres + Redis","：默认 SQLite 高吞吐崩",[42,2271,2272,2275],{},[45,2273,2274],{},"Worker 模式","：高并发要起 worker container 才能并行",[42,2277,2278,2281],{},[45,2279,2280],{},"Webhook 加防护","：公网 Webhook 加 IP 白名单 \u002F secret \u002F nginx",[42,2283,2284,2287],{},[45,2285,2286],{},"数据加密","：n8n encryption key 设强随机值，备份要带 key",[42,2289,2290,2293],{},[45,2291,2292],{},"Cloud vs Self-host 成本","：>2k execution\u002F月 自托管更省",[42,2295,2296,2299],{},[45,2297,2298],{},"集成缺失","：冷门 SaaS 用 HTTP Request + curl 等价",[42,2301,2302,2305],{},[45,2303,2304],{},"AI Agent 稳定性","：生产关键流先用线性节点，agent 留给探索任务",[42,2307,2308,2311],{},[45,2309,2310],{},"license 法务","：Sustainable Use License 给法务看一遍，企业内部用没问题",[42,2313,2314,2316],{},[45,2315,1838],{},"：导入前看作者 + star 数 + 不要直接生产用，要 review",[42,2318,2319,2322],{},[45,2320,2321],{},"monitoring","：生产部署加 prometheus + 错误告警",[24,2324,461],{"id":460},[39,2326,2327,2330,2333,2336,2339,2342,2345,2348],{},[42,2328,2329],{},"✅ 开发者 + 完全控制 + 高量级自动化",[42,2331,2332],{},"✅ 从 Zapier \u002F Make 迁出降本",[42,2334,2335],{},"✅ AI Agent + LangChain + MCP 一体",[42,2337,2338],{},"✅ 合规 \u002F 数据驻留 \u002F 自托管需求",[42,2340,2341],{},"❌ 非技术 + 要 8000+ 现成集成（用 Zapier）",[42,2343,2344],{},"❌ 完全不愿自托管 + 不想付 Cloud",[42,2346,2347],{},"❌ 纯研究 \u002F 学术 agent（用 OpenManus）",[42,2349,2350],{},"❌ 极简 2-step 自动化（Zapier 更快）",[24,2352,524],{"id":524},[39,2354,2355,2360,2364],{},[42,2356,2357],{},[528,2358,2359],{"href":1229},"Langflow 评测",[42,2361,2362],{},[528,2363,1154],{"href":1153},[42,2365,2366],{},[528,2367,1160],{"href":1159},[24,2369,542],{"id":542},[227,2371,2372,2379,2386,2393],{},[42,2373,2374,2375],{},"n8n 官网 ",[528,2376,2377],{"href":2377,"rel":2378},"https:\u002F\u002Fn8n.io",[556],[42,2380,2381,2382],{},"AutomationByExperts — n8n 2026 200k users 5x ARR ",[528,2383,2384],{"href":2384,"rel":2385},"https:\u002F\u002Fautomationbyexperts.com\u002Fblog\u002Fn8n-ai-workflow-automation-guide-2026",[556],[42,2387,2388,2389],{},"Tutorials Technology — n8n + AI on Linux 2026（Docker + Ollama）",[528,2390,2391],{"href":2391,"rel":2392},"https:\u002F\u002Ftutorials.technology\u002Ftutorials\u002Fn8n-ai-workflows-linux-2026.html",[556],[42,2394,2395,2396],{},"Northflank — n8n Self-host Architecture + Pricing 2026 ",[528,2397,2398],{"href":2398,"rel":2399},"https:\u002F\u002Fnorthflank.com\u002Fblog\u002Fhow-to-self-host-n8n-setup-architecture-and-pricing-guide",[556],[1194,2401,2402],{},"html pre.shiki code .sJ8bj, html code.shiki .sJ8bj{--shiki-default:#6A737D;--shiki-dark:#6A737D}html pre.shiki code .sScJk, html code.shiki .sScJk{--shiki-default:#6F42C1;--shiki-dark:#B392F0}html pre.shiki code .sj4cs, html code.shiki .sj4cs{--shiki-default:#005CC5;--shiki-dark:#79B8FF}html pre.shiki code .sZZnC, html code.shiki .sZZnC{--shiki-default:#032F62;--shiki-dark:#9ECBFF}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html.dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}",{"title":566,"searchDepth":567,"depth":567,"links":2404},[2405,2406,2407,2408,2409,2410,2411,2412,2413,2414],{"id":26,"depth":570,"text":27},{"id":37,"depth":570,"text":37},{"id":99,"depth":570,"text":99},{"id":1874,"depth":570,"text":1875},{"id":1938,"depth":570,"text":1939},{"id":270,"depth":570,"text":270},{"id":409,"depth":570,"text":409},{"id":460,"depth":570,"text":461},{"id":524,"depth":570,"text":524},{"id":542,"depth":570,"text":542},"\u002Fimg\u002Ftools\u002Fn8n.webp","n8n 2026 真实评测：开源自托管自动化平台和 AI Agent 工作流工具，支持 LangChain 节点、MCP、Ollama、OpenAI、Webhook、400+ 集成和 execution-based 定价。本文对比 Zapier、Make、Dify，整理自托管成本、适合场景和避坑建议。",[2418,2421,2424,2427],{"q":2419,"a":2420},"n8n 和 Zapier \u002F Make 怎么选？","Zapier 8000+ 集成 + 最易上手 + 非技术团队最爱，但 $29.99\u002F月才 750 tasks + 每步独立计费 = 量大成本爆炸。Make 视觉画布 + 并行分支 + 2000+ 集成 + 智能打包步骤（10k ops $29\u002F月），中等复杂最佳性价比。n8n 开源 + 自托管 + execution-based（步骤数无关）+ 70 LangChain 节点 + MCP 原生，开发者 + 高量级 + 完全控制首选。模式：从 Zapier \u002F Make 起步 → 撞墙 → 迁 n8n。",{"q":2422,"a":2423},"AI Agent 节点是什么？","n8n 2026 加的特殊节点：运行 reasoning loop，从连接的工具节点中自主挑选并调用，直到有答案。和传统线性节点的『按顺序执行预定动作』不同，AI Agent 引入了 LLM 决策。可挂接近 70 LangChain 节点 \u002F MCP server \u002F 任意 HTTP API。让 n8n 从『纯自动化』升级为『真正的 AI agent 平台』。",{"q":2425,"a":2426},"Sustainable Use License 是什么协议？","n8n 用的 fair-code 协议（非传统 OSI 开源）。允许内部使用 + 自托管 + 修改源码，但限制把 n8n 作为 SaaS 转售（与 n8n 商业版直接竞争）。对自用 \u002F 内部工具 \u002F 普通自托管 100% 免费。要做 n8n competitor \u002F 商业 SaaS 才需要谈授权。",{"q":2428,"a":2429},"自托管成本和门槛？","最小：$5-10\u002F月 VPS（DigitalOcean \u002F Hetzner \u002F Linode）+ Docker 一行起。Postgres 持久化 + Redis 队列（高吞吐）+ Worker 节点（水平扩展）。10 分钟内能跑通最小版本。中文社区 \u002F B 站 \u002F 知乎 有大量中文教程。比 Langflow \u002F Dify 上手快。","https:\u002F\u002Fgithub.com\u002Fn8n-io\u002Fn8n",[589,1226],{},[1231,1232,596,1233],[2435,2439,2443,2447],{"plan":2436,"price":129,"features":2437,"notes":2438},"Community (Self-host)","全部功能 + 不限执行 + 不限 workflow + Sustainable Use License","VPS $5-10\u002F月",{"plan":1854,"price":2440,"features":2441,"notes":2442},"~€20\u002F月","2,500 executions + 5 workflow + 基础集成","试水 \u002F 小团队",{"plan":1860,"price":2444,"features":2445,"notes":2446},"~€50\u002F月","更高 executions + 团队协作 + 高级特性","中型团队",{"plan":741,"price":150,"features":2448,"notes":2449},"SSO + audit + LDAP + 私有部署 + SLA","大客户","Self-host 免费 \u002F Cloud Starter ~€20·月 (2.5k executions) \u002F Pro \u002F Business \u002F Enterprise 阶梯",[2452],"onboarding\u002Fn8n-ollama-automation",{"power":601,"ux":600,"price":601,"cn_support":567,"stability":601},{"title":924,"description":2416},"n8n 评测 2026：开源自托管自动化平台，AI Agent 工作流首选",[2457,2459,2461,2463],{"name":2458,"url":2377,"accessed":1259},"n8n 官网",{"name":2460,"url":2384,"accessed":1259},"AutomationByExperts — n8n 2026 200k users 5x ARR",{"name":2462,"url":2391,"accessed":1259},"Tutorials Technology — n8n + AI on Linux 2026",{"name":2464,"url":2398,"accessed":1259},"Northflank — n8n Self-host Pricing 2026","tools\u002Fagent\u002Fplatform\u002Fn8n","自托管自动化平台：200k+ 用户 + 70 LangChain 节点 + MCP 原生 + Ollama 集成",[615,2468,2469,1270,2470,1231,924],"workflow","automation","mcp","Zapier \u002F Make 的开源替代——20 步工作流和 2 步成本一样（自托管）。AI Agent + LangChain 节点让它在 2026 成为开发者首选自动化平台。要 8000+ 现成集成 + 极简上手用 Zapier；要中等复杂 + 不自托管用 Make。","-iDDIKwjGjB9NT7Qxs4dB1VzNj1sd4TtsHpoVRTDw9A",1785660639249]