zdh
5167d9e36e
Deploy Wiki to Production / deploy (push) Has been cancelled
覆盖 27 个此前缺少返回链接的报告页和独立页面（共 50 个 HTML 页，index.html 首页除外） - 标准报告页：header/header-bar 内插入 back-link - 特殊页面：深国际、中医馆、YqBoot、交互式演示等单独适配 - 统一文案「← 返回知识库」，URL 按目录深度自动推算
1004 lines
73 KiB
HTML
1004 lines
73 KiB
HTML
<!DOCTYPE html>
|
||
<html lang="zh-CN">
|
||
<head>
|
||
<meta charset="UTF-8">
|
||
<meta name="viewport" content="width=device-width, initial-scale=1.0">
|
||
<title>全功能智能体平台深度对比 — AI Agent 五大系统调研报告</title>
|
||
<style>
|
||
:root {
|
||
--color-primary: #1677FF; --color-primary-hover: #4096FF; --color-primary-active: #0958D9; --color-primary-bg: #E6F4FF;
|
||
--color-success: #52C41A; --color-success-bg: #F6FFED; --color-success-border: #B7EB8F;
|
||
--color-warning: #FAAD14; --color-warning-bg: #FFFBE6; --color-warning-border: #FFE58F;
|
||
--color-error: #FF4D4F; --color-error-bg: #FFF2F0; --color-error-border: #FFCCC7;
|
||
--bg: #F5F5F5; --bg-container: #FFFFFF; --bg-elevated: #FAFAFA;
|
||
--border: #D9D9D9; --border-light: #F0F0F0;
|
||
--text-primary: #141414; --text-secondary: #595959; --text-tertiary: #8C8C8C;
|
||
--radius-sm: 6px; --radius-md: 8px; --radius-lg: 12px;
|
||
--shadow-sm: 0 1px 2px rgba(0,0,0,0.03), 0 1px 6px -1px rgba(0,0,0,0.02);
|
||
--shadow-md: 0 2px 4px rgba(0,0,0,0.04), 0 4px 12px -2px rgba(0,0,0,0.04);
|
||
--shadow-lg: 0 4px 8px rgba(0,0,0,0.06), 0 8px 24px -4px rgba(0,0,0,0.08);
|
||
--sidebar-w: 260px; --header-h: 64px;
|
||
}
|
||
*{margin:0;padding:0;box-sizing:border-box;}
|
||
body{font-family:-apple-system,BlinkMacSystemFont,'Segoe UI',Roboto,'PingFang SC','Hiragino Sans GB','Microsoft YaHei',sans-serif;background:var(--bg);color:var(--text-primary);line-height:1.57;font-size:14px;}
|
||
::-webkit-scrollbar{width:6px;height:6px;}
|
||
::-webkit-scrollbar-track{background:transparent;}
|
||
::-webkit-scrollbar-thumb{background:var(--border);border-radius:3px;}
|
||
|
||
.header{position:fixed;top:0;left:0;right:0;height:var(--header-h);background:rgba(255,255,255,0.88);backdrop-filter:blur(12px);border-bottom:1px solid var(--border-light);display:flex;align-items:center;padding:0 2rem;z-index:100;}
|
||
.header h1{font-size:18px;font-weight:700;background:linear-gradient(135deg,var(--color-primary),#722ED1);-webkit-background-clip:text;background-clip:text;-webkit-text-fill-color:transparent;color:transparent;}
|
||
.header .back-link{margin-right:16px;font-size:13px;color:var(--color-primary);text-decoration:none;flex-shrink:0;}
|
||
.header .back-link:hover{text-decoration:underline;}
|
||
.header .version{margin-left:auto;font-size:12px;color:var(--text-tertiary);font-family:'SF Mono',Monaco,monospace;}
|
||
|
||
.menu-btn{display:none;width:40px;height:40px;border:none;background:none;cursor:pointer;border-radius:var(--radius-sm);font-size:20px;margin-right:12px;flex-shrink:0;}
|
||
.menu-btn:hover{background:var(--bg-elevated);}
|
||
.drawer-overlay{display:none;position:fixed;inset:0;background:rgba(0,0,0,0.5);z-index:200;opacity:0;transition:opacity .25s;}
|
||
.drawer-overlay.show{display:block;opacity:1;}
|
||
|
||
.sidebar{position:fixed;top:var(--header-h);left:0;bottom:0;width:var(--sidebar-w);background:var(--bg-container);border-right:1px solid var(--border);overflow-y:auto;padding:1.25rem 0;z-index:90;}
|
||
.sidebar .toc-label{font-size:11px;text-transform:uppercase;letter-spacing:.1em;color:var(--text-tertiary);padding:0 1.25rem;margin-bottom:.75rem;font-weight:600;}
|
||
.sidebar .toc-part{font-size:12px;font-weight:700;color:var(--color-primary);padding:.6rem 1.25rem .2rem;margin-top:.4rem;}
|
||
.sidebar nav ol{list-style:none;counter-reset:toc;}
|
||
.sidebar nav li{counter-increment:toc;}
|
||
.sidebar nav a{display:block;padding:.4rem 1.25rem;color:var(--text-secondary);text-decoration:none;font-size:12px;line-height:1.4;border-left:2px solid transparent;transition:all .15s;}
|
||
.sidebar nav a:hover,.sidebar nav a.active{color:var(--color-primary);background:var(--color-primary-bg);}
|
||
.sidebar nav a.active{border-left-color:var(--color-primary);font-weight:600;}
|
||
|
||
.main{margin-left:var(--sidebar-w);margin-top:var(--header-h);padding:32px 40px 64px;min-height:calc(100vh - var(--header-h));max-width:1100px;}
|
||
section{margin:48px 0;scroll-margin-top:calc(var(--header-h) + 1rem);}
|
||
section h2{font-size:24px;line-height:32px;padding-bottom:8px;border-bottom:1px solid var(--border);margin-bottom:20px;color:var(--color-primary);}
|
||
section h3{font-size:20px;line-height:28px;margin:24px 0 12px;color:#722ED1;}
|
||
section h4{font-size:16px;line-height:24px;margin:16px 0 8px;color:var(--color-success);}
|
||
p{color:var(--text-secondary);font-size:14px;margin:10px 0;text-align:justify;}
|
||
|
||
blockquote{background:var(--color-primary-bg);border-left:4px solid var(--color-primary);margin:16px 0;padding:12px 20px;border-radius:0 var(--radius-sm) var(--radius-sm) 0;}
|
||
blockquote p{text-indent:0;color:var(--text-primary);font-size:14px;}
|
||
strong{color:var(--text-primary);}
|
||
|
||
.box{border-radius:var(--radius-md);padding:20px 24px;margin:18px 0;}
|
||
.box-info{background:var(--color-primary-bg);border:1px solid #91CAFF;}
|
||
.box-idea{background:#F9F0FF;border:1px solid #D3ADF7;}
|
||
.box-good{background:var(--color-success-bg);border:1px solid var(--color-success-border);}
|
||
.box-warn{background:var(--color-warning-bg);border:1px solid var(--color-warning-border);}
|
||
.box-danger{background:var(--color-error-bg);border:1px solid var(--color-error-border);}
|
||
.box h4{margin:0 0 8px;font-size:14px;font-weight:700;}
|
||
.box-info h4{color:var(--color-primary);}.box-idea h4{color:#722ED1;}.box-good h4{color:var(--color-success);}.box-warn h4{color:var(--color-warning);}.box-danger h4{color:var(--color-error);}
|
||
.box p{font-size:14px;margin:4px 0;text-indent:0;}
|
||
|
||
.table-wrap{overflow-x:auto;margin:20px 0;}
|
||
.table-caption{font-size:13px;font-weight:700;color:var(--text-primary);margin-bottom:8px;}
|
||
table{width:100%;border-collapse:collapse;font-size:13px;}
|
||
th,td{padding:10px 14px;border:1px solid var(--border);text-align:left;vertical-align:top;}
|
||
th{background:var(--bg-elevated);font-weight:700;color:var(--text-primary);white-space:nowrap;}
|
||
td{background:var(--bg-container);color:var(--text-secondary);}
|
||
tbody tr:hover td{background:var(--color-primary-bg);}
|
||
|
||
.tag{display:inline-block;padding:2px 10px;border-radius:12px;font-size:11px;font-weight:600;margin-right:4px;}
|
||
.tag-green{background:var(--color-success-bg);color:var(--color-success);border:1px solid var(--color-success-border);}
|
||
.tag-blue{background:var(--color-primary-bg);color:var(--color-primary);border:1px solid #91CAFF;}
|
||
.tag-warn{background:var(--color-warning-bg);color:var(--color-warning);border:1px solid var(--color-warning-border);}
|
||
.tag-red{background:var(--color-error-bg);color:var(--color-error);border:1px solid var(--color-error-border);}
|
||
.tag-purple{background:#F9F0FF;color:#722ED1;border:1px solid #D3ADF7;}
|
||
.tag-teal{background:#E6FFFB;color:#13C2C2;border:1px solid #87E8DE;}
|
||
|
||
.kpi-row{display:flex;flex-wrap:wrap;gap:12px;margin:16px 0;}
|
||
.kpi{flex:1;min-width:110px;background:var(--color-primary-bg);border-radius:var(--radius-md);padding:14px 12px;text-align:center;}
|
||
.kpi .num{font-size:20px;font-weight:800;color:var(--color-primary);line-height:1.2;}
|
||
.kpi .label{font-size:10px;color:var(--text-tertiary);margin-top:4px;}
|
||
.kpi.warn{background:var(--color-warning-bg);}.kpi.warn .num{color:#D48806;}
|
||
.kpi.good{background:var(--color-success-bg);}.kpi.good .num{color:var(--color-success);}
|
||
.kpi.danger{background:var(--color-error-bg);}.kpi.danger .num{color:var(--color-error);}
|
||
.kpi.purple{background:#F9F0FF;}.kpi.purple .num{color:#722ED1;}
|
||
|
||
ul,ol{margin:8px 0 8px 20px;}
|
||
li{margin:4px 0;color:var(--text-secondary);font-size:14px;}
|
||
|
||
.card-grid{display:grid;grid-template-columns:repeat(auto-fill,minmax(300px,1fr));gap:16px;margin:16px 0;}
|
||
.card{background:var(--bg-container);border:1px solid var(--border);border-radius:var(--radius-md);padding:20px 24px;transition:all .2s;box-shadow:var(--shadow-sm);}
|
||
.card:hover{border-color:var(--color-primary);box-shadow:var(--shadow-md);}
|
||
.card h4{font-size:15px;font-weight:700;margin:0 0 4px;color:var(--text-primary);}
|
||
.card .card-meta{font-size:11px;color:var(--text-tertiary);margin-bottom:6px;}
|
||
.card .card-desc{font-size:13px;color:var(--text-secondary);line-height:1.6;}
|
||
.card .card-tags{display:flex;flex-wrap:wrap;gap:4px;margin-top:8px;}
|
||
|
||
.pros-cons{display:grid;grid-template-columns:1fr 1fr;gap:16px;margin:12px 0;}
|
||
.pros{background:var(--color-success-bg);border:1px solid var(--color-success-border);border-radius:var(--radius-md);padding:16px 20px;}
|
||
.cons{background:var(--color-error-bg);border:1px solid var(--color-error-border);border-radius:var(--radius-md);padding:16px 20px;}
|
||
.pros h4{color:var(--color-success);margin:0 0 8px;}
|
||
.cons h4{color:var(--color-error);margin:0 0 8px;}
|
||
.pros ul,.cons ul{padding-left:18px;font-size:13px;}
|
||
.pros li,.cons li{margin-bottom:2px;}
|
||
|
||
.footer{text-align:center;padding:28px 0;color:var(--text-tertiary);font-size:13px;border-top:1px solid var(--border-light);margin-top:40px;}
|
||
|
||
.back-to-top{display:none;position:fixed;bottom:24px;right:24px;width:48px;height:48px;border-radius:50%;background:var(--color-primary);color:#fff;border:none;cursor:pointer;font-size:20px;box-shadow:var(--shadow-md);z-index:80;}
|
||
.back-to-top:hover{transform:scale(1.1);}
|
||
.back-to-top.show{display:flex;align-items:center;justify-content:center;}
|
||
|
||
.star{color:#D48806;font-weight:700;}
|
||
.rank-1 td{background:#FFFBE6!important;}
|
||
.rank-2 td{background:#F6FFED!important;}
|
||
.rank-3 td{background:#E6F4FF!important;}
|
||
|
||
@media(max-width:992px){:root{--sidebar-w:220px;}.main{padding:24px 20px 48px;}}
|
||
@media(max-width:768px){
|
||
.menu-btn{display:flex;align-items:center;justify-content:center;}
|
||
.sidebar{display:block;position:fixed;top:0;left:0;bottom:0;width:280px;max-width:85vw;transform:translateX(-100%);transition:transform .25s ease;z-index:201;padding-top:60px;}
|
||
.drawer-overlay.show ~ .sidebar,.drawer-overlay.show .sidebar,.sidebar.show{transform:translateX(0);}
|
||
.main{margin-left:0;padding:20px 16px 40px;}
|
||
.header h1{font-size:15px;}
|
||
.pros-cons{grid-template-columns:1fr;}
|
||
.card-grid{grid-template-columns:1fr;}
|
||
}
|
||
</style>
|
||
</head>
|
||
<body>
|
||
|
||
<div class="drawer-overlay" id="drawerOverlay"></div>
|
||
|
||
<header class="header">
|
||
<button class="menu-btn" id="menuBtn" aria-label="打开目录">☰</button>
|
||
<a href="../../index.html" class="back-link">← 返回知识库</a>
|
||
<h1>全功能智能体平台深度对比</h1>
|
||
<span class="version">V1.0 · 2026-06-04</span>
|
||
</header>
|
||
|
||
<aside class="sidebar">
|
||
<div class="toc-label">报告目录</div>
|
||
<nav>
|
||
<div class="toc-part">总览</div>
|
||
<ol>
|
||
<li><a href="#s0">调研概览：10 大核心平台</a></li>
|
||
<li><a href="#s0-1">五大功能覆盖度矩阵</a></li>
|
||
</ol>
|
||
<div class="toc-part">平台逐一详解</div>
|
||
<ol>
|
||
<li><a href="#s1">Hermes Agent — "自进化"标杆</a></li>
|
||
<li><a href="#s2">OpenClaw — "平台之王"</a></li>
|
||
<li><a href="#s3">Spring AI Alibaba — "最均衡企业级平台"</a></li>
|
||
<li><a href="#s4">其他平台速览（SemaClaw · Dify · AnimaWorks · KiwiQ · CowAgent · GoClaw · Gitee Xtreme）</a></li>
|
||
</ol>
|
||
<div class="toc-part">横向对比</div>
|
||
<ol>
|
||
<li><a href="#s5">开源协议与商用分析</a></li>
|
||
<li><a href="#s6">开源版 vs 商业版差异</a></li>
|
||
<li><a href="#s7">优缺点对比总表</a></li>
|
||
<li><a href="#s8">技术栈对比</a></li>
|
||
<li><a href="#s9">GitHub vs Gitee 生态差异</a></li>
|
||
</ol>
|
||
<div class="toc-part">生态索引</div>
|
||
<ol>
|
||
<li><a href="#s10">五大子系统开源项目索引</a></li>
|
||
<li><a href="#s11">选型建议</a></li>
|
||
<li><a href="#s12">开源地址速查（国内镜像）</a></li>
|
||
</ol>
|
||
</nav>
|
||
</aside>
|
||
|
||
<main class="main">
|
||
|
||
<!-- ============================================================ -->
|
||
<!-- 总览 -->
|
||
<!-- ============================================================ -->
|
||
<section id="s0">
|
||
<h2>调研概览：10 大核心平台</h2>
|
||
|
||
<blockquote><p>筛选标准：覆盖<strong>多智能体平面 · 低代码编排 · Harness基建 · 技能市场 · 智能体评测</strong>五类中至少三类的平台。调研范围涵盖 GitHub（7 个）+ Gitee（3 个），共 10 个核心平台，关联收录 81 个开源项目。</p></blockquote>
|
||
|
||
<div class="kpi-row">
|
||
<div class="kpi"><div class="num">10</div><div class="label">核心平台</div></div>
|
||
<div class="kpi good"><div class="num">81</div><div class="label">收录项目</div></div>
|
||
<div class="kpi warn"><div class="num">5</div><div class="label">功能维度</div></div>
|
||
<div class="kpi purple"><div class="num">63</div><div class="label">GitHub 项目</div></div>
|
||
<div class="kpi"><div class="num">18</div><div class="label">Gitee 项目</div></div>
|
||
</div>
|
||
|
||
<div class="card-grid">
|
||
<div class="card">
|
||
<h4>🔥 Hermes Agent</h4>
|
||
<div class="card-meta">GitHub <span class="star">★ 140K+</span> · MIT · Nous Research</div>
|
||
<div class="card-desc">自进化AI助手。三层记忆(SQLite+FTS5)+自动Skill生成+自我进化闭环。2026年2月开源，两个月内增长最快项目之一。</div>
|
||
<div class="card-tags"><span class="tag tag-blue">自进化</span><span class="tag tag-purple">三层记忆</span><span class="tag tag-green">自动Skill</span><span class="tag tag-warn">200+模型</span></div>
|
||
</div>
|
||
<div class="card">
|
||
<h4>🦞 OpenClaw</h4>
|
||
<div class="card-meta">GitHub <span class="star">★ 355K+</span> · Apache 2.0</div>
|
||
<div class="card-desc">Gateway-First微内核架构。25+IM通道，物理多Agent隔离，ClawHub社区 44K+技能市场，本地优先+五层安全防御。</div>
|
||
<div class="card-tags"><span class="tag tag-blue">多Agent网关</span><span class="tag tag-green">44K技能</span><span class="tag tag-purple">25+通道</span><span class="tag tag-warn">本地优先</span></div>
|
||
</div>
|
||
<div class="card">
|
||
<h4>☕ Spring AI Alibaba</h4>
|
||
<div class="card-meta">GitHub/Gitee <span class="star">★ 20K+</span> · Apache 2.0 · 阿里巴巴</div>
|
||
<div class="card-desc">唯一企业级Java Agent框架。Graph工作流引擎+Admin可视化+Nacos服务注册+A2A分布式通信，五项能力最均衡。</div>
|
||
<div class="card-tags"><span class="tag tag-blue">Java企业级</span><span class="tag tag-purple">Graph引擎</span><span class="tag tag-green">Admin平台</span><span class="tag tag-warn">A2A</span></div>
|
||
</div>
|
||
<div class="card">
|
||
<h4>🔧 SemaClaw</h4>
|
||
<div class="card-meta">GitHub 新项目 · MIT · 美的AI团队</div>
|
||
<div class="card-desc">四层插件架构(MCP工具/子代理/技能/钩子)+DAG Teams混合编排+Workbench交付物渲染。Git仓库即插件源。</div>
|
||
<div class="card-tags"><span class="tag tag-purple">插件架构</span><span class="tag tag-blue">DAG编排</span><span class="tag tag-green">Workbench</span></div>
|
||
</div>
|
||
<div class="card">
|
||
<h4>📐 Dify</h4>
|
||
<div class="card-meta">GitHub <span class="star">★ 80K+</span> · Apache 2.0 · LangGenius</div>
|
||
<div class="card-desc">可视化低代码行业标杆。50+官方工具，RAG管道最成熟，Apache 2.0全开源+云服务双轨。多Agent编排能力有限。</div>
|
||
<div class="card-tags"><span class="tag tag-green">可视化标杆</span><span class="tag tag-blue">RAG成熟</span><span class="tag tag-purple">零代码</span></div>
|
||
</div>
|
||
<div class="card">
|
||
<h4>🧠 AnimaWorks</h4>
|
||
<div class="card-meta">GitHub 新项目 · MIT</div>
|
||
<div class="card-desc">组织级多Agent架构。神经科学启发三阶段遗忘记忆，6通道自动启动，心跳24/7自主运行，组织层级+委托审计。</div>
|
||
<div class="card-tags"><span class="tag tag-purple">神经科学记忆</span><span class="tag tag-blue">组织层级</span><span class="tag tag-warn">24/7自主</span></div>
|
||
</div>
|
||
<div class="card">
|
||
<h4>⚡ KiwiQ</h4>
|
||
<div class="card-meta">GitHub 新项目 · MIT</div>
|
||
<div class="card-desc">200+企业Agent实战验证后全开源。JSON定义Agent+27+预置生产工作流+内置可观测性。</div>
|
||
<div class="card-tags"><span class="tag tag-green">企业验证</span><span class="tag tag-blue">27+工作流</span><span class="tag tag-purple">可观测</span></div>
|
||
</div>
|
||
<div class="card">
|
||
<h4>🐮 CowAgent</h4>
|
||
<div class="card-meta">Gitee · 开源 · 国产</div>
|
||
<div class="card-desc">国产超级AI助理。三级记忆+Deep Dream夜间蒸馏+MEMORY.md长期记忆+多IM渠道(微信/飞书/钉钉)。</div>
|
||
<div class="card-tags"><span class="tag tag-purple">Deep Dream</span><span class="tag tag-blue">三级记忆</span><span class="tag tag-green">多IM</span></div>
|
||
</div>
|
||
<div class="card">
|
||
<h4>🔷 GoClaw</h4>
|
||
<div class="card-meta">Gitee · CC BY-NC 4.0</div>
|
||
<div class="card-desc">Go语言多Agent网关。8阶段Pipeline+3层记忆+4种Prompt模式+多租户+自我进化，桌面版Wails+React UI。</div>
|
||
<div class="card-tags"><span class="tag tag-blue">Go高性能</span><span class="tag tag-purple">8阶段Pipeline</span><span class="tag tag-green">多租户</span></div>
|
||
</div>
|
||
<div class="card">
|
||
<h4>🏗️ Gitee Xtreme AI</h4>
|
||
<div class="card-meta">Gitee 官方 · 部分开源</div>
|
||
<div class="card-desc">Gitee DevOps原生集成。多Agent并行协作+MCP协议统一调度+完整上下文工程体系+数据飞轮评测闭环。</div>
|
||
<div class="card-tags"><span class="tag tag-blue">DevOps原生</span><span class="tag tag-purple">MCP调度</span><span class="tag tag-green">数据飞轮</span></div>
|
||
</div>
|
||
</div>
|
||
</section>
|
||
|
||
<!-- ============================================================ -->
|
||
<!-- 覆盖度矩阵 -->
|
||
<!-- ============================================================ -->
|
||
<section id="s0-1">
|
||
<h2>五大功能覆盖度矩阵</h2>
|
||
|
||
<blockquote><p>评分：<span class="tag tag-green">⭐⭐⭐⭐⭐</span> 业界领先 <span class="tag tag-blue">⭐⭐⭐⭐</span> 完善 <span class="tag tag-warn">⭐⭐⭐</span> 及格 <span class="tag tag-red">⭐⭐</span> 基础 ⭐ 缺失</p></blockquote>
|
||
|
||
<div class="table-wrap">
|
||
<div class="table-caption">表：10 大平台 × 5 维功能覆盖度矩阵</div>
|
||
<table>
|
||
<thead><tr><th>平台</th><th style="text-align:center">多智能体平面</th><th style="text-align:center">低代码编排</th><th style="text-align:center">Harness基建</th><th style="text-align:center">技能市场</th><th style="text-align:center">评测服务</th><th style="text-align:center">总分/25</th><th>排名</th></tr></thead>
|
||
<tbody>
|
||
<tr class="rank-1">
|
||
<td><strong>Spring AI Alibaba</strong></td>
|
||
<td style="text-align:center"><span class="tag tag-green">5</span></td>
|
||
<td style="text-align:center"><span class="tag tag-blue">4</span></td>
|
||
<td style="text-align:center"><span class="tag tag-blue">4</span></td>
|
||
<td style="text-align:center"><span class="tag tag-warn">3</span></td>
|
||
<td style="text-align:center"><span class="tag tag-blue">4</span></td>
|
||
<td style="text-align:center;font-weight:800">21</td>
|
||
<td><span class="tag tag-green">#1</span></td>
|
||
</tr>
|
||
<tr class="rank-2">
|
||
<td><strong>OpenClaw</strong></td>
|
||
<td style="text-align:center"><span class="tag tag-green">5</span></td>
|
||
<td style="text-align:center"><span class="tag tag-red">2</span></td>
|
||
<td style="text-align:center"><span class="tag tag-blue">4</span></td>
|
||
<td style="text-align:center"><span class="tag tag-green">5</span></td>
|
||
<td style="text-align:center"><span class="tag tag-red">2</span></td>
|
||
<td style="text-align:center;font-weight:800">18</td>
|
||
<td><span class="tag tag-blue">#2</span></td>
|
||
</tr>
|
||
<tr class="rank-3">
|
||
<td><strong>Hermes Agent</strong></td>
|
||
<td style="text-align:center"><span class="tag tag-warn">3</span></td>
|
||
<td style="text-align:center"><span class="tag tag-red">2</span></td>
|
||
<td style="text-align:center"><span class="tag tag-green">5</span></td>
|
||
<td style="text-align:center"><span class="tag tag-blue">4</span></td>
|
||
<td style="text-align:center"><span class="tag tag-red">2</span></td>
|
||
<td style="text-align:center;font-weight:800">17</td>
|
||
<td><span class="tag tag-purple">#3</span></td>
|
||
</tr>
|
||
<tr><td><strong>GoClaw</strong></td><td style="text-align:center"><span class="tag tag-blue">4</span></td><td style="text-align:center"><span class="tag tag-red">2</span></td><td style="text-align:center"><span class="tag tag-blue">4</span></td><td style="text-align:center"><span class="tag tag-red">2</span></td><td style="text-align:center"><span class="tag tag-red">2</span></td><td style="text-align:center;font-weight:800">15</td><td>#4</td></tr>
|
||
<tr><td><strong>SemaClaw</strong></td><td style="text-align:center"><span class="tag tag-warn">3</span></td><td style="text-align:center"><span class="tag tag-warn">3</span></td><td style="text-align:center"><span class="tag tag-warn">3</span></td><td style="text-align:center"><span class="tag tag-blue">4</span></td><td style="text-align:center"><span class="tag tag-red">2</span></td><td style="text-align:center;font-weight:800">15</td><td>#5</td></tr>
|
||
<tr><td><strong>Dify</strong></td><td style="text-align:center"><span class="tag tag-red">2</span></td><td style="text-align:center"><span class="tag tag-green">5</span></td><td style="text-align:center"><span class="tag tag-warn">3</span></td><td style="text-align:center"><span class="tag tag-warn">3</span></td><td style="text-align:center"><span class="tag tag-warn">3</span></td><td style="text-align:center;font-weight:800">15</td><td>#6</td></tr>
|
||
<tr><td><strong>Gitee Xtreme AI</strong></td><td style="text-align:center"><span class="tag tag-blue">4</span></td><td style="text-align:center"><span class="tag tag-red">2</span></td><td style="text-align:center"><span class="tag tag-blue">4</span></td><td style="text-align:center"><span class="tag tag-red">2</span></td><td style="text-align:center"><span class="tag tag-warn">3</span></td><td style="text-align:center;font-weight:800">15</td><td>#7</td></tr>
|
||
<tr><td><strong>KiwiQ</strong></td><td style="text-align:center"><span class="tag tag-blue">4</span></td><td style="text-align:center"><span class="tag tag-red">2</span></td><td style="text-align:center"><span class="tag tag-warn">3</span></td><td style="text-align:center"><span class="tag tag-warn">3</span></td><td style="text-align:center"><span class="tag tag-warn">3</span></td><td style="text-align:center;font-weight:800">14</td><td>#8</td></tr>
|
||
<tr><td><strong>AnimaWorks</strong></td><td style="text-align:center"><span class="tag tag-blue">4</span></td><td style="text-align:center">⭐</td><td style="text-align:center"><span class="tag tag-blue">4</span></td><td style="text-align:center">⭐</td><td style="text-align:center">⭐</td><td style="text-align:center;font-weight:800">13</td><td>#9</td></tr>
|
||
<tr><td><strong>CowAgent</strong></td><td style="text-align:center"><span class="tag tag-red">2</span></td><td style="text-align:center">⭐</td><td style="text-align:center"><span class="tag tag-green">5</span></td><td style="text-align:center"><span class="tag tag-warn">3</span></td><td style="text-align:center">⭐</td><td style="text-align:center;font-weight:800">12</td><td>#10</td></tr>
|
||
</tbody>
|
||
</table>
|
||
</div>
|
||
|
||
<div class="box box-warn">
|
||
<h4>⚠️ 关键发现</h4>
|
||
<p>目前<strong>没有一个平台在五项功能上全部达到 ⭐⭐⭐⭐ 以上</strong>。最接近的是 Spring AI Alibaba（21/25），但在技能市场方面仍需加强。2026年的最佳实践是<strong>两个平台组合使用，通过 MCP 协议互联互通</strong>。</p>
|
||
</div>
|
||
</section>
|
||
|
||
<!-- ============================================================ -->
|
||
<!-- S1: Hermes Agent -->
|
||
<!-- ============================================================ -->
|
||
<section id="s1">
|
||
<h2>Hermes Agent — "自进化"标杆</h2>
|
||
|
||
<blockquote><p>Nous Research 出品，2026年2月开源，发布后两个月内突破 140K Stars。核心定位：会<strong>自我进化（Self-Improving）</strong>的 AI Agent，拥有跨会话持久记忆、自动技能生成与优化闭环。</p></blockquote>
|
||
|
||
<h3>核心架构：五大支柱</h3>
|
||
|
||
<div class="table-wrap">
|
||
<div class="table-caption">表：Hermes Agent 五层架构</div>
|
||
<table>
|
||
<thead><tr><th style="width:12%">支柱</th><th style="width:18%">名称</th><th>说明</th></tr></thead>
|
||
<tbody>
|
||
<tr><td><strong>支柱1</strong></td><td>三层记忆系统</td><td>Tier1: MEMORY.md + USER.md（持久核心，严格字符上限）→ Tier2: SQLite+FTS5全文索引（跨会话搜索+LLM摘要）→ Tier3: 8个插件式外部记忆</td></tr>
|
||
<tr><td><strong>支柱2</strong></td><td>Skill 系统</td><td>40+预置 + 520+社区Hub + Agent主动创建。Tool调用超5次自动提炼为可复用Skill，发现缺陷自动patch修补</td></tr>
|
||
<tr><td><strong>支柱3</strong></td><td>SOUL.md 身份层</td><td>定义Agent性格/语气/沟通风格和硬性边界，永远位于system prompt第一位，支持交互式进化</td></tr>
|
||
<tr><td><strong>支柱4</strong></td><td>Cron 自然语言调度</td><td>"每天早上8点搜索最新论文"→自动生成Skill+Cron Job，Cron会话自成隔离</td></tr>
|
||
<tr><td><strong>支柱5</strong></td><td>自我进化闭环</td><td>记忆策划→Skill创建→Skill自改进→跨会话FTS5召回→Nudge静默审查，五步闭环全自动</td></tr>
|
||
</tbody>
|
||
</table>
|
||
</div>
|
||
|
||
<div class="pros-cons">
|
||
<div class="pros"><h4>✅ 核心优势</h4><ul>
|
||
<li>越用越聪明：记忆+Skill 复利效应</li>
|
||
<li>开箱即用安全：沙箱+授权审批+安全扫描</li>
|
||
<li>200+模型支持（OpenRouter/本地 vLLM）</li>
|
||
<li>6种执行环境 + 6平台消息网关</li>
|
||
<li>记忆快照冻结——保护前缀缓存、节省Token</li>
|
||
</ul></div>
|
||
<div class="cons"><h4>❌ 核心劣势</h4><ul>
|
||
<li>冷启动期：新部署时记忆/Skill为空</li>
|
||
<li>反馈质量决定进化效果，自动≠魔法</li>
|
||
<li>非企业级服务治理（无注册中心/熔断）</li>
|
||
<li>记忆污染：过期记忆可能导致行为异常</li>
|
||
<li>仍偏开发者工具，有配置门槛</li>
|
||
</ul></div>
|
||
</div>
|
||
</section>
|
||
|
||
<!-- ============================================================ -->
|
||
<!-- S2: OpenClaw -->
|
||
<!-- ============================================================ -->
|
||
<section id="s2">
|
||
<h2>OpenClaw — "平台之王"</h2>
|
||
|
||
<blockquote><p>社区驱动，355K+ Stars，全球最大Agent社区。Gateway-First微内核架构，Local-First隐私保护，物理多Agent隔离。ClawHub社区 44,000+ Skills。</p></blockquote>
|
||
|
||
<h3>五层架构</h3>
|
||
|
||
<div class="table-wrap">
|
||
<div class="table-caption">表：OpenClaw 五层架构</div>
|
||
<table>
|
||
<thead><tr><th style="width:14%">层</th><th style="width:18%">名称</th><th>说明</th></tr></thead>
|
||
<tbody>
|
||
<tr><td><strong>L1</strong></td><td>Channel 层</td><td>25+ IM 适配器（Telegram/Discord/QQ/飞书/WhatsApp/Signal/微信）</td></tr>
|
||
<tr><td><strong>L2</strong></td><td>Orchestration 层</td><td>Gateway 路由+Agent 调度+认证+配置热重载，9级路由优先级</td></tr>
|
||
<tr><td><strong>L3</strong></td><td>Capability 层</td><td>Plugins & Skills SDK，4种协作模式（Supervisor/Router/Pipeline/Parallel）</td></tr>
|
||
<tr><td><strong>L4</strong></td><td>Memory 层</td><td>LanceDB向量+Knowledge Graph+Active Recall+后台Dreaming整合</td></tr>
|
||
<tr><td><strong>L5</strong></td><td>Model 层</td><td>9 LLM Provider + 多模型故障转移链</td></tr>
|
||
</tbody>
|
||
</table>
|
||
</div>
|
||
|
||
<div class="pros-cons">
|
||
<div class="pros"><h4>✅ 核心优势</h4><ul>
|
||
<li>平台广度无与伦比（25+通道/44K技能）</li>
|
||
<li>本地优先隐私保护，数据不离设备</li>
|
||
<li>物理隔离防上下文污染——每个Agent独立workspace</li>
|
||
<li>五层纵深安全防御（TLS→Device ID→Approval→Sandbox→Scanning）</li>
|
||
<li>跨平台会话迁移（独特能力）</li>
|
||
</ul></div>
|
||
<div class="cons"><h4>❌ 核心劣势</h4><ul>
|
||
<li>Token消耗大（单轮可达数千Token）</li>
|
||
<li>默认压缩有损，长会话丢失上下文</li>
|
||
<li>复杂多步任务易丢失关键决策</li>
|
||
<li>ClawHub供应链风险（曾报告API Key窃取）</li>
|
||
<li>学习曲线陡峭（需理解MCP/编排/技能开发）</li>
|
||
</ul></div>
|
||
</div>
|
||
</section>
|
||
|
||
<!-- ============================================================ -->
|
||
<!-- S3: Spring AI Alibaba -->
|
||
<!-- ============================================================ -->
|
||
<section id="s3">
|
||
<h2>Spring AI Alibaba — "最均衡企业级平台"</h2>
|
||
|
||
<blockquote><p>阿里巴巴出品，Apache 2.0。唯一企业级Java Agent框架。三层架构：Agent Framework → Graph Runtime → Augmented LLM。五种功能最均衡，总分 21/25。</p></blockquote>
|
||
|
||
<h3>核心能力</h3>
|
||
|
||
<div class="table-wrap">
|
||
<div class="table-caption">表：Spring AI Alibaba 核心能力矩阵</div>
|
||
<table>
|
||
<thead><tr><th style="width:14%">能力域</th><th style="width:16%">评分</th><th>关键实现</th></tr></thead>
|
||
<tbody>
|
||
<tr><td><strong>多智能体</strong></td><td><span class="tag tag-green">5</span></td><td>6种编排模式（Sequential/Parallel/Routing/Loop/Supervisor/Handoff）+ Nacos服务注册 + A2A跨服务通信</td></tr>
|
||
<tr><td><strong>低代码编排</strong></td><td><span class="tag tag-blue">4</span></td><td>Admin可视化拖拽 + 低/高/零代码三种模式 + PlantUML/Mermaid导出 + Graph引擎条件路由/嵌套/中断恢复</td></tr>
|
||
<tr><td><strong>Harness基建</strong></td><td><span class="tag tag-blue">4</span></td><td>上下文压缩+编辑+持久化 + Human-in-the-Loop + Agent Skills渐进式披露（大幅节省Token）</td></tr>
|
||
<tr><td><strong>技能市场</strong></td><td><span class="tag tag-warn">3</span></td><td>Agent Skills机制（v1.1.2.0新增），渐进式披露降低Token，无公开社区技能市场</td></tr>
|
||
<tr><td><strong>评测服务</strong></td><td><span class="tag tag-blue">4</span></td><td>Admin平台内置：数据集管理+评估器+实验管理+结果分析 + OpenTelemetry全链路追踪</td></tr>
|
||
</tbody>
|
||
</table>
|
||
</div>
|
||
|
||
<div class="pros-cons">
|
||
<div class="pros"><h4>✅ 核心优势</h4><ul>
|
||
<li>Java企业级生态，Spring开发者零门槛</li>
|
||
<li>Graph引擎工作流能力业界领先</li>
|
||
<li>Admin平台实现 开发→编排→评估→监控 标准化流水线</li>
|
||
<li>A2A+Nacos原生企业级服务治理</li>
|
||
<li>国产化+阿里云原生集成+评测内置</li>
|
||
</ul></div>
|
||
<div class="cons"><h4>❌ 核心劣势</h4><ul>
|
||
<li>语言锁定Java，Python AI生态不兼容</li>
|
||
<li>GitHub Star较低(20K vs 355K)，国际影响力弱</li>
|
||
<li>技能市场不成熟，无社区贡献生态</li>
|
||
<li>部署复杂（需Nacos/数据库/Admin等多组件）</li>
|
||
<li>开源版功能可能弱于阿里云商业版</li>
|
||
</ul></div>
|
||
</div>
|
||
</section>
|
||
|
||
<!-- ============================================================ -->
|
||
<!-- S4: 其他平台 -->
|
||
<!-- ============================================================ -->
|
||
<section id="s4">
|
||
<h2>其他平台速览</h2>
|
||
|
||
<div class="table-wrap">
|
||
<div class="table-caption">表：SemaClaw · Dify · AnimaWorks · KiwiQ · CowAgent · GoClaw · Gitee Xtreme AI 七平台速览</div>
|
||
<table>
|
||
<thead><tr><th style="width:10%">平台</th><th style="width:12%">最强项</th><th style="width:12%">最弱项</th><th>一句话定位</th><th>最佳场景</th></tr></thead>
|
||
<tbody>
|
||
<tr><td><strong>SemaClaw</strong></td><td>技能市场（4）</td><td>评测（2）</td><td>四层插件架构 + Git仓库即插件市场</td><td>插件化Agent开发；交付物生成</td></tr>
|
||
<tr><td><strong>Dify</strong></td><td>低代码（5）</td><td>多Agent（2）</td><td>可视化低代码行业标杆</td><td>RAG应用；聊天机器人；快速原型</td></tr>
|
||
<tr><td><strong>AnimaWorks</strong></td><td>Harness（4）</td><td>编排/市场/评测</td><td>"代码即组织"，神经科学记忆</td><td>研究型Agent；长时组织协作</td></tr>
|
||
<tr><td><strong>KiwiQ</strong></td><td>多Agent（4）</td><td>编排（2）</td><td>200+企业验证后全开源</td><td>生产级Agent工作流</td></tr>
|
||
<tr><td><strong>CowAgent</strong></td><td>Harness（5）</td><td>编排/评测</td><td>Deep Dream夜间蒸馏+多IM</td><td>个人AI助理；微信/钉钉集成</td></tr>
|
||
<tr><td><strong>GoClaw</strong></td><td>多Agent+Harness</td><td>技能/评测</td><td>Go语言8阶段Pipeline网关</td><td>Go技术栈Agent网关</td></tr>
|
||
<tr><td><strong>Gitee Xtreme AI</strong></td><td>多Agent+Harness</td><td>编排/技能</td><td>DevOps原生+MCP统一调度</td><td>Gitee用户研发智能化</td></tr>
|
||
</tbody>
|
||
</table>
|
||
</div>
|
||
</section>
|
||
|
||
<!-- ============================================================ -->
|
||
<!-- S5: 开源协议与商用分析 -->
|
||
<!-- ============================================================ -->
|
||
<section id="s5">
|
||
<h2>开源协议与商用分析</h2>
|
||
|
||
<blockquote><p>开源不等于可商用。以下逐一分析五个核心平台的<strong>开源协议、商用限制、以及开源版与商业版的真实差异</strong>。</p></blockquote>
|
||
|
||
<div class="table-wrap">
|
||
<div class="table-caption">表：五大平台开源协议与商用许可速查</div>
|
||
<table>
|
||
<thead><tr><th style="width:10%">平台</th><th style="width:16%">开源协议</th><th style="width:10%">可商用</th><th>商用条件与限制</th><th style="width:8%">商业版</th></tr></thead>
|
||
<tbody>
|
||
<tr>
|
||
<td><strong>Hermes Agent</strong></td>
|
||
<td><span class="tag tag-green">MIT</span></td>
|
||
<td><span class="tag tag-green">✅ 可</span></td>
|
||
<td>MIT 最宽松协议，允许任意使用、修改、分发、商用。需保留版权声明。无任何限制。</td>
|
||
<td><span class="tag tag-red">无</span></td>
|
||
</tr>
|
||
<tr>
|
||
<td><strong>OpenClaw</strong></td>
|
||
<td><span class="tag tag-green">Apache 2.0</span></td>
|
||
<td><span class="tag tag-green">✅ 可</span></td>
|
||
<td>Apache 2.0 允许商用、修改、分发。需保留版权声明+许可声明+修改说明。专利授权条款保护用户不受专利诉讼。核心项目完全开源免费，但国内厂商推出多种衍生商业版（见下文）。</td>
|
||
<td><span class="tag tag-green">有</span></td>
|
||
</tr>
|
||
<tr>
|
||
<td><strong>Spring AI Alibaba</strong></td>
|
||
<td><span class="tag tag-green">Apache 2.0</span></td>
|
||
<td><span class="tag tag-green">✅ 可</span></td>
|
||
<td>同 Apache 2.0，核心框架完全开源可商用。但部分高级能力（ARMS可观测、百炼平台集成）依赖阿里云商业服务，需阿里云账号并产生云资源费用。无"付费企业版"限制，云服务按量计费。</td>
|
||
<td><span class="tag tag-warn">云服务</span></td>
|
||
</tr>
|
||
<tr>
|
||
<td><strong>SemaClaw</strong></td>
|
||
<td><span class="tag tag-green">MIT</span></td>
|
||
<td><span class="tag tag-green">✅ 可</span></td>
|
||
<td>MIT 协议，允许任意使用。项目较新，目前仅开源社区版，无商业版。</td>
|
||
<td><span class="tag tag-red">无</span></td>
|
||
</tr>
|
||
<tr>
|
||
<td><strong>Dify</strong></td>
|
||
<td><span class="tag tag-warn">Apache 2.0<br>+ 附加限制</span></td>
|
||
<td><span class="tag tag-warn">⚠️ 有条件</span></td>
|
||
<td><strong>修改版 Apache 2.0</strong>，含两条额外限制：① <strong>禁止多租户商用</strong>——不能用开源版搭建 SaaS 服务卖给外部客户；② <strong>禁止去 Logo</strong>——不可移除 Dify 控制台和应用中的 Logo。<br>内部使用完全免费。对外销售 SaaS 必须购买企业版商业授权。</td>
|
||
<td><span class="tag tag-green">有</span></td>
|
||
</tr>
|
||
</tbody>
|
||
</table>
|
||
</div>
|
||
|
||
<div class="box box-warn">
|
||
<h4>⚠️ 商用风险提示</h4>
|
||
<p><strong>Dify 是唯一有明确商用限制的平台。</strong>如果用 Dify 开源版搭建 SaaS 服务对外收费，可能面临法律风险。其他四个平台（MIT / Apache 2.0）均可自由商用，但需注意：</p>
|
||
<p>1. <strong>MIT</strong>：最宽松，无任何附加条件，但无专利保护条款</p>
|
||
<p>2. <strong>Apache 2.0</strong>：含专利授权，对商用更友好（防止贡献者后续发起专利诉讼）</p>
|
||
<p>3. 所有协议均<strong>不提供担保</strong>——使用开源软件的商业风险由使用者自行承担</p>
|
||
</div>
|
||
|
||
</section>
|
||
|
||
<!-- ============================================================ -->
|
||
<!-- S6: 开源版 vs 商业版差异 -->
|
||
<!-- ============================================================ -->
|
||
<section id="s6">
|
||
<h2>开源版 vs 商业版差异详解</h2>
|
||
|
||
<blockquote><p>部分平台存在"开源社区版 + 商业增强版"双轨制。以下逐平台分析<strong>开源版的实际功能边界、缺失的企业级特性、以及商业版提供什么额外价值</strong>。</p></blockquote>
|
||
|
||
<!-- Hermes Agent -->
|
||
<h3>🔥 Hermes Agent — 纯开源，无商业版</h3>
|
||
|
||
<div class="table-wrap">
|
||
<div class="table-caption">表：Hermes Agent 开源版功能完整度</div>
|
||
<table>
|
||
<thead><tr><th style="width:14%">维度</th><th>开源版（唯一版本）</th><th>局限性</th></tr></thead>
|
||
<tbody>
|
||
<tr><td><strong>部署方式</strong></td><td>CLI 安装，本地部署</td><td>无 Windows 原生支持（需 WSL2）；国内需代理</td></tr>
|
||
<tr><td><strong>记忆系统</strong></td><td>三层记忆（SQLite+FTS5+外部插件），全功能开放</td><td>30天后知识图谱可能产生 +17% 冗余；需自行维护压缩</td></tr>
|
||
<tr><td><strong>Skill 系统</strong></td><td>40+预置 + 520+社区 + 自动生成，全功能开放</td><td>社区 Skill 质量参差不齐，需自行审查</td></tr>
|
||
<tr><td><strong>安全</strong></td><td>五层安全（TLS/沙箱/扫描）</td><td>无 RBAC、无审计日志、无合规认证（ISO 27001等）</td></tr>
|
||
<tr><td><strong>高可用</strong></td><td>单实例运行</td><td>不支持集群、无自动扩缩容、无 SLA 保障</td></tr>
|
||
<tr><td><strong>模型</strong></td><td>200+模型，需自备 API Key</td><td>API 费用自理；长期运行 6个月后延迟可能 +22%</td></tr>
|
||
<tr><td><strong>运维</strong></td><td>需 AI 工程团队自行维护</td><td>需 2-4 周学习+配置；持续维护成本高</td></tr>
|
||
<tr><td><strong>企业级支持</strong></td><td>社区支持</td><td>无 SLA、无专属技术支持、无定制开发服务</td></tr>
|
||
</tbody>
|
||
</table>
|
||
</div>
|
||
|
||
<div class="box box-info">
|
||
<h4>Hermes Agent 开源版适合谁？</h4>
|
||
<p>技术能力强、有运维团队的开发者或小团队。如果需要企业级的 RBAC、审计、合规、SLA 保障，目前只能<strong>自行二次开发</strong>。Nous Research 未发布官方商业版，但有社区提及的 RDSHermes 等第三方商业包装版本（增加审计+加密托管+WebUI，价格未公开）。</p>
|
||
</div>
|
||
|
||
<!-- OpenClaw -->
|
||
<h3>🦞 OpenClaw — 开源纯净版 vs 国内厂商商业版</h3>
|
||
|
||
<div class="table-wrap">
|
||
<div class="table-caption">表：OpenClaw 开源版 vs 国内厂商商业版</div>
|
||
<table>
|
||
<thead><tr><th style="width:16%">维度</th><th>开源版（OpenClaw）</th><th>国内厂商商业版</th></tr></thead>
|
||
<tbody>
|
||
<tr><td><strong>厂商</strong></td><td>社区维护</td><td>阿里云 HiClaw · 腾讯云 ClawPro · 中兴 Co-Claw · 中关村科金 PowerClaw · 月之暗面 KimiClaw</td></tr>
|
||
<tr><td><strong>部署</strong></td><td>本地单机（Mac/Windows/Linux）</td><td>私有云 / VPC / 混合云，支持 1000+ 用户多租户</td></tr>
|
||
<tr><td><strong>权限管控</strong></td><td>❌ 无 RBAC，用户可调用任何工具</td><td>✅ 细粒度 RBAC + 时间/IP/设备约束</td></tr>
|
||
<tr><td><strong>审计追踪</strong></td><td>❌ 无审计日志</td><td>✅ 全链路审计：每次工具调用、API 调用、数据访问可追溯</td></tr>
|
||
<tr><td><strong>审批流程</strong></td><td>❌ 不支持</td><td>✅ 敏感操作需人工审批</td></tr>
|
||
<tr><td><strong>安全隔离</strong></td><td>基础社区级沙箱</td><td>✅ 多层安全沙箱 + 操作拦截 + 五层安全模型</td></tr>
|
||
<tr><td><strong>合规认证</strong></td><td>❌ 无</td><td>✅ SOC 2 / ISO 27001 / HIPAA / 等保</td></tr>
|
||
<tr><td><strong>SLA</strong></td><td>❌ 社区支持</td><td>✅ 99.9% SLA（如腾讯云 ClawPro）</td></tr>
|
||
<tr><td><strong>通道集成</strong></td><td>25+ 通道（需手动配置）</td><td>预集成企业微信/钉钉/飞书/QQ，开箱即用</td></tr>
|
||
<tr><td><strong>技能市场</strong></td><td>ClawHub 44K+（社区，12% 含恶意代码）</td><td>企业审核技能市场，质量可控</td></tr>
|
||
<tr><td><strong>价格</strong></td><td>免费</td><td>按坐席/按量/订阅制（通常 ¥2万-10万+/年）</td></tr>
|
||
</tbody>
|
||
</table>
|
||
</div>
|
||
|
||
<div class="box box-danger">
|
||
<h4>⚠️ 重要安全注意</h4>
|
||
<p>2026年3月，<strong>香港数字政策办公室明确指示：OpenClaw 不得安装在政府内网</strong>，因其权限范围大且缺乏审计能力。CrowdStrike 报告称 ClawHub 社区技能中 <strong>12% 含恶意代码</strong>。企业用户不建议直接使用开源版处理敏感数据。</p>
|
||
</div>
|
||
|
||
<!-- Spring AI Alibaba -->
|
||
<h3>☕ Spring AI Alibaba — 开源框架 + 阿里云服务</h3>
|
||
|
||
<div class="table-wrap">
|
||
<div class="table-caption">表：Spring AI Alibaba 开源能力 vs 依赖阿里云的部分</div>
|
||
<table>
|
||
<thead><tr><th style="width:16%">能力</th><th>开源版（免费可用）</th><th>需阿里云商业服务</th></tr></thead>
|
||
<tbody>
|
||
<tr><td><strong>多Agent编排</strong></td><td>✅ Graph框架：6种编排模式，完整开放</td><td>—</td></tr>
|
||
<tr><td><strong>工作流引擎</strong></td><td>✅ DAG + 条件路由 + 中断恢复，完整开放</td><td>—</td></tr>
|
||
<tr><td><strong>Admin 可视化</strong></td><td>✅ Studio 调试 + 数据集管理 + 评估器，完整开放</td><td>—</td></tr>
|
||
<tr><td><strong>MCP 分布式</strong></td><td>✅ Nacos MCP Registry，开源可用</td><td>—</td></tr>
|
||
<tr><td><strong>提示词管理</strong></td><td>✅ Nacos 动态配置，开源可用</td><td>—</td></tr>
|
||
<tr><td><strong>可观测性</strong></td><td>基础 OpenTelemetry 埋点</td><td>🔗 ARMS 深度集成（阿里云商业产品）</td></tr>
|
||
<tr><td><strong>模型服务</strong></td><td>兼容 OpenAI API，可接任何模型</td><td>🔗 百炼平台（通义千问/万相，按 Token 计费）</td></tr>
|
||
<tr><td><strong>AI 网关</strong></td><td>自建</td><td>🔗 Higress AI 网关（阿里云商业产品）</td></tr>
|
||
<tr><td><strong>对象存储</strong></td><td>兼容 S3</td><td>🔗 阿里云 OSS（按量计费）</td></tr>
|
||
<tr><td><strong>Serverless</strong></td><td>自建</td><td>🔗 阿里云函数计算（按调用计费）</td></tr>
|
||
<tr><td><strong>商业支持</strong></td><td>社区 GitHub Issues</td><td>🔗 阿里云企业支持计划（SLA保障）</td></tr>
|
||
</tbody>
|
||
</table>
|
||
</div>
|
||
|
||
<div class="box box-idea">
|
||
<h4>💡 关键认知</h4>
|
||
<p>Spring AI Alibaba <strong>没有"付费解锁功能"的概念</strong>——所有 Agent 编排、工作流、Admin 平台在开源版中已完整提供。差异在于：开源版是一个<strong>框架</strong>（你自己搭基础设施），阿里云商业服务提供的是<strong>托管能力</strong>（监控、网关、模型服务等开箱即用）。你可以完全用开源版 + 自建基础设施（如 Prometheus+Grafana 替代 ARMS），成本更低但需要运维投入。</p>
|
||
</div>
|
||
|
||
<!-- SemaClaw -->
|
||
<h3>🔧 SemaClaw — 纯开源，新兴项目</h3>
|
||
|
||
<div class="table-wrap">
|
||
<div class="table-caption">表：SemaClaw 开源版分析</div>
|
||
<table>
|
||
<thead><tr><th style="width:14%">维度</th><th>现状</th><th>说明</th></tr></thead>
|
||
<tbody>
|
||
<tr><td><strong>协议</strong></td><td><span class="tag tag-green">MIT</span></td><td>最宽松，允许任意使用和商用</td></tr>
|
||
<tr><td><strong>商业版</strong></td><td><span class="tag tag-red">无</span></td><td>美的 AI 团队开源，目前无商业版计划公开</td></tr>
|
||
<tr><td><strong>成熟度</strong></td><td><span class="tag tag-warn">早期</span></td><td>2026年新项目，社区规模小，文档不完善</td></tr>
|
||
<tr><td><strong>核心能力</strong></td><td>四层插件架构 + DAG Teams + Workbench</td><td>架构设计创新，但功能覆盖度不如成熟平台</td></tr>
|
||
<tr><td><strong>企业就绪</strong></td><td><span class="tag tag-red">否</span></td><td>无 RBAC/审计/合规/SLA/高可用</td></tr>
|
||
</tbody>
|
||
</table>
|
||
</div>
|
||
|
||
<!-- Dify -->
|
||
<h3>📐 Dify — 开源版 vs 云端版 vs 企业版</h3>
|
||
|
||
<div class="table-wrap">
|
||
<div class="table-caption">表：Dify 三版本详细对比（2026年定价）</div>
|
||
<table>
|
||
<thead><tr><th style="width:14%">维度</th><th>开源社区版</th><th>云端 Pro ($59/月)</th><th>云端 Team ($159/月)</th><th>企业版 (定制)</th></tr></thead>
|
||
<tbody>
|
||
<tr><td><strong>部署</strong></td><td>自托管 Docker/K8s</td><td>SaaS（AWS 美国区）</td><td>SaaS（AWS 美国区）</td><td>私有化 / 专属云</td></tr>
|
||
<tr><td><strong>应用数</strong></td><td>无限制</td><td>50</td><td>200</td><td>定制</td></tr>
|
||
<tr><td><strong>消息额度</strong></td><td>无限制（仅受模型限制）</td><td>5,000/月</td><td>10,000/月</td><td>定制</td></tr>
|
||
<tr><td><strong>团队席位</strong></td><td>无限制</td><td>3 人</td><td>50 人</td><td>不限</td></tr>
|
||
<tr><td><strong>知识库</strong></td><td>无限制</td><td>5GB</td><td>20GB</td><td>定制</td></tr>
|
||
<tr><td><strong>自定义插件</strong></td><td><span class="tag tag-green">✅</span></td><td><span class="tag tag-red">❌</span></td><td><span class="tag tag-red">❌</span></td><td><span class="tag tag-green">✅</span></td></tr>
|
||
<tr><td><strong>私有模型接入</strong></td><td><span class="tag tag-green">✅</span> 任何 OpenAI 兼容 API</td><td><span class="tag tag-red">❌</span> 仅限官方支持</td><td><span class="tag tag-red">❌</span></td><td><span class="tag tag-green">✅</span></td></tr>
|
||
<tr><td><strong>源码修改</strong></td><td><span class="tag tag-green">✅</span> 完全可魔改</td><td><span class="tag tag-red">❌</span> 黑盒</td><td><span class="tag tag-red">❌</span></td><td>部分可定制</td></tr>
|
||
<tr><td><strong>多租户商用</strong></td><td><span class="tag tag-red">❌ 禁止</span></td><td><span class="tag tag-green">✅</span></td><td><span class="tag tag-green">✅</span></td><td><span class="tag tag-green">✅</span> 含商业授权</td></tr>
|
||
<tr><td><strong>去 Logo</strong></td><td><span class="tag tag-red">❌ 禁止</span></td><td><span class="tag tag-red">❌</span></td><td><span class="tag tag-red">❌</span></td><td><span class="tag tag-green">✅</span> 唯一合法途径</td></tr>
|
||
<tr><td><strong>SSO/RBAC</strong></td><td><span class="tag tag-red">❌</span></td><td><span class="tag tag-red">❌</span></td><td><span class="tag tag-red">❌</span></td><td><span class="tag tag-green">✅</span></td></tr>
|
||
<tr><td><strong>SLA</strong></td><td>社区</td><td>邮件优先</td><td>邮件优先</td><td>专属 SLA</td></tr>
|
||
<tr><td><strong>等保/合规</strong></td><td>可满足（自部署）</td><td><span class="tag tag-red">❌</span> 海外服务器</td><td><span class="tag tag-red">❌</span></td><td><span class="tag tag-green">✅</span> SOC2 Type II</td></tr>
|
||
</tbody>
|
||
</table>
|
||
</div>
|
||
|
||
<div class="box box-warn">
|
||
<h4>⚠️ Dify 国内容器部署的隐性成本</h4>
|
||
<p>云端版服务器在 AWS 美国区，<strong>国内模型厂商（阿里百炼、智谱等）的新用户免费额度几乎无法使用</strong>（境外 IP 调用被拒绝或按商用计费）。自托管开源版本地直连国产模型 API，可 100% 享受免费额度，模型成本可接近零。建议国内用户优先自托管。</p>
|
||
</div>
|
||
|
||
<!-- 总结对比 -->
|
||
<h3>五平台开源 vs 商业总览</h3>
|
||
|
||
<div class="table-wrap">
|
||
<table>
|
||
<thead><tr><th style="width:10%">平台</th><th>开源版完整度</th><th>商业版核心增量</th><th>商用建议</th></tr></thead>
|
||
<tbody>
|
||
<tr>
|
||
<td><strong>Hermes Agent</strong></td>
|
||
<td><span class="tag tag-green">100%</span></td>
|
||
<td>无官方商业版。第三方 RDSHermes 增加审计+加密+WebUI</td>
|
||
<td>个人/小团队直接使用；企业需二次开发 RBAC/审计</td>
|
||
</tr>
|
||
<tr>
|
||
<td><strong>OpenClaw</strong></td>
|
||
<td><span class="tag tag-blue">核心框架完整</span></td>
|
||
<td>RBAC/审计/合规/SLA/企业集成/技能审核——国内厂商提供全套企业级能力</td>
|
||
<td>个人免费使用；企业建议用国内厂商商业版（阿里/腾讯/中兴）</td>
|
||
</tr>
|
||
<tr>
|
||
<td><strong>Spring AI Alibaba</strong></td>
|
||
<td><span class="tag tag-green">95%</span></td>
|
||
<td>阿里云托管服务（ARMS/百炼/Higress/OSS/FC）降低运维成本</td>
|
||
<td>框架层 100% 免费商用；云服务按需选用，可自建替代</td>
|
||
</tr>
|
||
<tr>
|
||
<td><strong>SemaClaw</strong></td>
|
||
<td><span class="tag tag-green">100%</span></td>
|
||
<td>无商业版。项目早期，体系不完整</td>
|
||
<td>关注即可，暂不建议生产使用</td>
|
||
</tr>
|
||
<tr>
|
||
<td><strong>Dify</strong></td>
|
||
<td><span class="tag tag-warn">功能完整但商用受限</span></td>
|
||
<td>去Logo/多租户商用授权/SSO/RBAC/SLA——企业版唯一合法途径</td>
|
||
<td>内部使用免费；对外 SaaS 必须购买企业版（¥万级/年）</td>
|
||
</tr>
|
||
</tbody>
|
||
</table>
|
||
</div>
|
||
|
||
</section>
|
||
|
||
<!-- ============================================================ -->
|
||
<!-- S7: 优缺点总表 -->
|
||
<!-- ============================================================ -->
|
||
<section id="s7">
|
||
<h2>优缺点对比总表</h2>
|
||
|
||
<div class="table-wrap">
|
||
<table>
|
||
<thead><tr><th>平台</th><th>核心优势</th><th>核心劣势</th><th>最佳场景</th></tr></thead>
|
||
<tbody>
|
||
<tr><td><strong>Hermes Agent</strong></td><td>自进化闭环+三层记忆+自动Skill生成</td><td>无服务治理/评测；冷启动期长</td><td>个人长期AI助理；需持续学习的场景</td></tr>
|
||
<tr><td><strong>OpenClaw</strong></td><td>355K社区+44K技能+25通道+物理Agent隔离</td><td>Token消耗大；上下文丢失；供应链风险</td><td>多平台自动化；企业多系统集成</td></tr>
|
||
<tr><td><strong>Spring AI Alibaba</strong></td><td>最均衡+Java生态+企业级治理+Admin平台</td><td>语言锁定Java；国际影响力弱；部署复杂</td><td>Java企业级Agent平台；国内企业</td></tr>
|
||
<tr><td><strong>SemaClaw</strong></td><td>Git仓库即插件源；Workbench交付物渲染</td><td>项目新；社区小；记忆系统弱</td><td>插件化Agent开发；交付物生成</td></tr>
|
||
<tr><td><strong>Dify</strong></td><td>可视化行业标杆+RAG成熟+零代码</td><td>多Agent天花板低；无真持久记忆</td><td>RAG应用；聊天机器人；快速原型</td></tr>
|
||
<tr><td><strong>AnimaWorks</strong></td><td>神经科学三阶段遗忘；组织层级架构</td><td>项目极小；多项功能缺失</td><td>研究型Agent；长时组织协作</td></tr>
|
||
<tr><td><strong>KiwiQ</strong></td><td>200+企业验证；27+生产工作流</td><td>无前端；社区极小</td><td>生产级Agent工作流</td></tr>
|
||
<tr><td><strong>CowAgent</strong></td><td>Deep Dream蒸馏；三级记忆；多IM</td><td>非多Agent；无评测；无编排</td><td>个人AI助理；微信/钉钉集成</td></tr>
|
||
<tr><td><strong>GoClaw</strong></td><td>Go高性能；8阶段Pipeline；多租户</td><td>社区极小；协议限制商用</td><td>Go技术栈Agent网关</td></tr>
|
||
<tr><td><strong>Gitee Xtreme AI</strong></td><td>DevOps原生集成；数据飞轮闭环</td><td>部分闭源；绑定Gitee生态</td><td>Gitee用户研发智能化</td></tr>
|
||
</tbody>
|
||
</table>
|
||
</div>
|
||
</section>
|
||
|
||
<!-- ============================================================ -->
|
||
<!-- S8: 技术栈对比 -->
|
||
<!-- ============================================================ -->
|
||
<section id="s8">
|
||
<h2>技术栈对比</h2>
|
||
|
||
<div class="table-wrap">
|
||
<div class="table-caption">表：10 平台技术栈详细对比</div>
|
||
<table>
|
||
<thead><tr><th>平台</th><th>语言</th><th>记忆存储</th><th>向量DB</th><th>通信协议</th><th>Agent编排</th><th>前端UI</th></tr></thead>
|
||
<tbody>
|
||
<tr><td><strong>Hermes Agent</strong></td><td>Python</td><td>SQLite+FTS5</td><td>—</td><td>MCP</td><td>Profile多实例</td><td>CLI</td></tr>
|
||
<tr><td><strong>OpenClaw</strong></td><td>TypeScript/Node</td><td>LanceDB+SQLite</td><td>LanceDB</td><td>MCP</td><td>Gateway路由</td><td>CLI</td></tr>
|
||
<tr><td><strong>Spring AI Alibaba</strong></td><td>Java 21+</td><td>对象存储</td><td>多支持</td><td>MCP+A2A+Nacos</td><td>Graph DAG</td><td>Admin (React)</td></tr>
|
||
<tr><td><strong>SemaClaw</strong></td><td>TypeScript/Node</td><td>Agentic Wiki</td><td>—</td><td>MCP</td><td>DAG Teams</td><td>Workbench</td></tr>
|
||
<tr><td><strong>Dify</strong></td><td>Python/Flask</td><td>PostgreSQL</td><td>多支持</td><td>REST/OpenAPI</td><td>可视化节点</td><td>React/Next.js</td></tr>
|
||
<tr><td><strong>AnimaWorks</strong></td><td>Python</td><td>Chroma+图DB</td><td>Chroma</td><td>SDK直连</td><td>层级组织</td><td>—</td></tr>
|
||
<tr><td><strong>KiwiQ</strong></td><td>Python</td><td>PostgreSQL</td><td>—</td><td>JSON</td><td>流水线</td><td>—</td></tr>
|
||
<tr><td><strong>CowAgent</strong></td><td>Python</td><td>本地文件+DB</td><td>—</td><td>MCP</td><td>单Agent</td><td>Gradio</td></tr>
|
||
<tr><td><strong>GoClaw</strong></td><td>Go</td><td>PostgreSQL+pgvector</td><td>pgvector</td><td>事件驱动</td><td>Agent Pipeline</td><td>Wails+React</td></tr>
|
||
<tr><td><strong>Gitee Xtreme AI</strong></td><td>Java/Go</td><td>—</td><td>—</td><td>MCP</td><td>并行协作</td><td>Web</td></tr>
|
||
</tbody>
|
||
</table>
|
||
</div>
|
||
|
||
<div class="box box-info">
|
||
<h4>关键技术趋势</h4>
|
||
<p>MCP协议 <strong>10/10 平台全部支持</strong> · 向量数据库 LanceDB/pgvector/Chroma 三足鼎立 · 记忆存储 SQLite+FTS5（轻量）/ PostgreSQL（企业）· 前端 React/ReactFlow 主流</p>
|
||
</div>
|
||
</section>
|
||
|
||
<!-- ============================================================ -->
|
||
<!-- S9: GitHub vs Gitee -->
|
||
<!-- ============================================================ -->
|
||
<section id="s9">
|
||
<h2>GitHub vs Gitee 生态差异</h2>
|
||
|
||
<div class="table-wrap">
|
||
<div class="table-caption">表：两大生态 10 维度横向对比</div>
|
||
<table>
|
||
<thead><tr><th style="width:14%">对比维度</th><th>GitHub 生态</th><th>Gitee 生态</th></tr></thead>
|
||
<tbody>
|
||
<tr><td><strong>社区规模</strong></td><td><span class="tag tag-green">极大规模</span> OpenClaw 355K⭐ / Hermes 140K⭐</td><td>规模较小：Spring AI Alibaba 20K⭐ 领跑</td></tr>
|
||
<tr><td><strong>标准化程度</strong></td><td><span class="tag tag-green">主导标准</span> SKILL.md / MCP / A2A / OTel</td><td>跟随国际标准 + MCP协议适配</td></tr>
|
||
<tr><td><strong>企业Java生态</strong></td><td><span class="tag tag-red">几乎空白</span></td><td><span class="tag tag-green">独特优势</span> Spring AI Alibaba 唯一企业级Java方案</td></tr>
|
||
<tr><td><strong>安全合规</strong></td><td>开源社区自发（如OpenClaw五层防御）</td><td><span class="tag tag-green">企业级</span> CVE强制扫描+哈希签名+细粒度权限</td></tr>
|
||
<tr><td><strong>评测体系</strong></td><td><span class="tag tag-blue">成熟</span> Future AGI/Opik/EvalMonkey 独立工具链</td><td><span class="tag tag-red">极弱</span> 仅 witty-skill-insight Skill专项，无通用评测</td></tr>
|
||
<tr><td><strong>链路追踪</strong></td><td><span class="tag tag-blue">标准化</span> OpenTelemetry 成为事实标准</td><td><span class="tag tag-red">空白</span> 无独立链路追踪开源项目</td></tr>
|
||
<tr><td><strong>技能市场</strong></td><td><span class="tag tag-blue">爆发增长</span> 150K+ Stars, ClawHub 44K+技能</td><td>起步期：Gitee Repo Skill仓库（企业内网定位）</td></tr>
|
||
<tr><td><strong>学术创新</strong></td><td><span class="tag tag-blue">极强</span> ACE/PRISM/KYA/Succession均有arXiv论文</td><td><span class="tag tag-red">弱</span> 以工程封装为主，缺乏原创研究</td></tr>
|
||
<tr><td><strong>中文友好</strong></td><td>仅少数项目有中文文档</td><td><span class="tag tag-green">全中文</span></td></tr>
|
||
<tr><td><strong>国产模型适配</strong></td><td>需自行适配</td><td><span class="tag tag-green">原生支持</span> 通义千问/DeepSeek/智谱等</td></tr>
|
||
</tbody>
|
||
</table>
|
||
</div>
|
||
</section>
|
||
|
||
<!-- ============================================================ -->
|
||
<!-- S8: 开源项目索引 -->
|
||
<!-- ============================================================ -->
|
||
<section id="s10">
|
||
<h2>五大子系统开源项目索引</h2>
|
||
|
||
<blockquote><p>除 10 个核心平台外，另收录各功能域的专项开源项目，共计 <strong>81 个</strong>（GitHub 63 个 + Gitee 18 个）。</p></blockquote>
|
||
|
||
<div class="table-wrap">
|
||
<div class="table-caption">表：按分类索引</div>
|
||
<table>
|
||
<thead><tr><th style="width:12%">分类</th><th style="width:8%">数量</th><th>重点推荐项目</th></tr></thead>
|
||
<tbody>
|
||
<tr><td><strong>多智能体平面</strong></td><td>14</td><td>agent-mesh (YAML注册+断路器) · OAN (DID/VC信任层) · KYA (策略合成) · CrewAI (72K⭐) · AutoGen (85K⭐) · AgentScope (45K⭐)</td></tr>
|
||
<tr><td><strong>低代码编排</strong></td><td>14</td><td>Open Agent Builder (拖拽+8节点) · Open Gumloop (Composio数百工具) · PaiAgent (Java) · Glink Engine (零依赖) · n8n (400+节点) · RuoYi AI · Conllect-LLM</td></tr>
|
||
<tr><td><strong>Harness基建</strong></td><td>22</td><td><strong>记忆：</strong>mem0 (48K⭐) · Hindsight (91.4% LongMemEval) · Letta/MemGPT (21K⭐) · Zep/Graphiti (时态KG) · GBrain · <strong>上下文：</strong>ACE (斯坦福自进化) · PRISM (2K→83.1%) · <strong>会话：</strong>Google ADK · CowAgent · LangGraphChatBot</td></tr>
|
||
<tr><td><strong>技能市场</strong></td><td>9</td><td>SkillNet (500K+技能) · skill-factory (300K+技能) · dora (9.5K内置) · SuperSkill · awesome-agent-skills · witty-skill-insight</td></tr>
|
||
<tr><td><strong>智能体评测</strong></td><td>11</td><td>Future AGI (50+指标+OTel) · Opik/Comet (18.7K⭐) · Agent Health/OpenSearch · agentevals/Solo.io · EvalMonkey (19基准+23混沌) · Langfuse (24.4K⭐)</td></tr>
|
||
</tbody>
|
||
</table>
|
||
</div>
|
||
</section>
|
||
|
||
<!-- ============================================================ -->
|
||
<!-- S11: 选型建议 -->
|
||
<!-- ============================================================ -->
|
||
<section id="s11">
|
||
<h2>选型建议</h2>
|
||
|
||
<div class="table-wrap">
|
||
<div class="table-caption">表：10 种场景推荐方案</div>
|
||
<table>
|
||
<thead><tr><th style="width:20%">场景</th><th>推荐方案</th><th>理由</th></tr></thead>
|
||
<tbody>
|
||
<tr><td><strong>个人：追求自进化+长期记忆</strong></td><td>Hermes Agent</td><td>三层记忆+自动Skill生成+自我进化闭环，记忆复利效应</td></tr>
|
||
<tr><td><strong>个人：多平台自动化</strong></td><td>OpenClaw</td><td>355K社区+44K技能+25通道，跨平台会话迁移</td></tr>
|
||
<tr><td><strong>个人：中文+微信/钉钉</strong></td><td>CowAgent</td><td>Deep Dream蒸馏+多IM原生集成+三级记忆</td></tr>
|
||
<tr><td><strong>企业：Java技术栈</strong></td><td>Spring AI Alibaba</td><td>唯一企业级Java方案+Graph引擎+Admin平台+A2A+Nacos</td></tr>
|
||
<tr><td><strong>企业：多系统集成</strong></td><td>OpenClaw</td><td>25+通道+物理Agent隔离+五层安全+ClawHub市场</td></tr>
|
||
<tr><td><strong>企业：DevOps研发智能化</strong></td><td>Gitee Xtreme AI</td><td>DevOps原生集成+MCP调度+数据飞轮闭环</td></tr>
|
||
<tr><td><strong>快速开发：RAG+聊天机器人</strong></td><td>Dify</td><td>可视化行业标杆+零代码+RAG管道最成熟</td></tr>
|
||
<tr><td><strong>快速开发：插件化Agent</strong></td><td>SemaClaw</td><td>Git仓库即插件源+四层插件架构+Workbench</td></tr>
|
||
<tr><td><strong>评测体系搭建</strong></td><td>Future AGI + EvalMonkey</td><td>50+评测指标+19基准+23混沌注入+OTel追踪</td></tr>
|
||
<tr><td><strong>记忆系统选型</strong></td><td>Hindsight / Mem0 / Zep</td><td>Hindsight 91.4%最高准确率；Mem0最快集成；Zep时态知识图谱</td></tr>
|
||
</tbody>
|
||
</table>
|
||
</div>
|
||
|
||
<div class="box box-idea">
|
||
<h4>💡 2026年最佳实践：组合使用，MCP 互联</h4>
|
||
<p><strong>个人用户：</strong>Hermes Agent（记忆+进化）+ OpenClaw（技能+多平台）</p>
|
||
<p><strong>Java 企业：</strong>Spring AI Alibaba（全栈底座）+ OpenClaw/Hermes（社区技能补充）</p>
|
||
<p><strong>国内企业：</strong>Spring AI Alibaba（企业底座）+ Gitee Xtreme AI（DevOps集成）</p>
|
||
<p><strong>快速验证：</strong>Dify（可视化原型）→ Spring AI Alibaba（生产落地）</p>
|
||
</div>
|
||
|
||
</section>
|
||
|
||
<!-- ============================================================ -->
|
||
<!-- S10: 开源地址速查（国内镜像） -->
|
||
<!-- ============================================================ -->
|
||
<section id="s12">
|
||
<h2>开源地址速查（国内镜像）</h2>
|
||
|
||
<blockquote><p>以下链接均已替换为国内可访问的镜像地址。GitHub 项目使用 <strong>gitclone.com</strong>（支持在线浏览+克隆），Gitee 项目使用原始地址。<br>如需克隆，可将 <code>gitclone.com</code> 替换为 <code>gitclone.com/github.com</code> 格式直接 <code>git clone</code>。</p></blockquote>
|
||
|
||
<div class="table-wrap">
|
||
<div class="table-caption">表：10 大核心平台</div>
|
||
<table>
|
||
<thead><tr><th style="width:14%">平台</th><th style="width:12%">来源</th><th>镜像地址（国内可访问）</th></tr></thead>
|
||
<tbody>
|
||
<tr><td><strong>Hermes Agent</strong></td><td>GitHub</td><td><a href="https://gitclone.com/github.com/NousResearch/hermes-agent" target="_blank">gitclone.com/github.com/NousResearch/hermes-agent</a></td></tr>
|
||
<tr><td><strong>OpenClaw</strong></td><td>GitHub</td><td><a href="https://gitclone.com/github.com/openclaw/openclaw" target="_blank">gitclone.com/github.com/openclaw/openclaw</a></td></tr>
|
||
<tr><td><strong>Spring AI Alibaba</strong></td><td>GitHub</td><td><a href="https://gitclone.com/github.com/alibaba/spring-ai-alibaba" target="_blank">gitclone.com/github.com/alibaba/spring-ai-alibaba</a></td></tr>
|
||
<tr><td><strong>Spring AI Alibaba</strong></td><td>Gitee</td><td><a href="https://gitee.com/alibaba/spring-ai-alibaba" target="_blank">gitee.com/alibaba/spring-ai-alibaba</a></td></tr>
|
||
<tr><td><strong>SemaClaw</strong></td><td>GitHub</td><td><a href="https://gitclone.com/github.com/nicepkg/semaclaw" target="_blank">gitclone.com/github.com/nicepkg/semaclaw</a></td></tr>
|
||
<tr><td><strong>Dify</strong></td><td>GitHub</td><td><a href="https://gitclone.com/github.com/langgenius/dify" target="_blank">gitclone.com/github.com/langgenius/dify</a></td></tr>
|
||
<tr><td><strong>AnimaWorks</strong></td><td>GitHub</td><td><a href="https://gitclone.com/github.com/xuiltul/animaworks" target="_blank">gitclone.com/github.com/xuiltul/animaworks</a></td></tr>
|
||
<tr><td><strong>KiwiQ</strong></td><td>GitHub</td><td><a href="https://gitclone.com/github.com/rcortx/kiwiq" target="_blank">gitclone.com/github.com/rcortx/kiwiq</a></td></tr>
|
||
<tr><td><strong>CowAgent</strong></td><td>Gitee</td><td><a href="https://gitee.com/zhayujie/CowAgent" target="_blank">gitee.com/zhayujie/CowAgent</a></td></tr>
|
||
<tr><td><strong>GoClaw</strong></td><td>Gitee</td><td><a href="https://gitee.com/devai/goclaw" target="_blank">gitee.com/devai/goclaw</a></td></tr>
|
||
</tbody>
|
||
</table>
|
||
</div>
|
||
|
||
<div class="table-wrap">
|
||
<div class="table-caption">表：多智能体平面专项项目</div>
|
||
<table>
|
||
<thead><tr><th style="width:14%">项目</th><th style="width:12%">来源</th><th>镜像地址</th></tr></thead>
|
||
<tbody>
|
||
<tr><td><strong>agent-mesh</strong></td><td>GitHub</td><td><a href="https://gitclone.com/github.com/reaatech/agent-mesh" target="_blank">gitclone.com/github.com/reaatech/agent-mesh</a></td></tr>
|
||
<tr><td><strong>OAN (OpenAgenet)</strong></td><td>GitHub</td><td><a href="https://gitclone.com/github.com/OpenAgenet" target="_blank">gitclone.com/github.com/OpenAgenet</a></td></tr>
|
||
<tr><td><strong>CrewAI</strong></td><td>GitHub</td><td><a href="https://gitclone.com/github.com/crewAIInc/crewAI" target="_blank">gitclone.com/github.com/crewAIInc/crewAI</a></td></tr>
|
||
<tr><td><strong>AutoGen (微软)</strong></td><td>GitHub</td><td><a href="https://gitclone.com/github.com/microsoft/autogen" target="_blank">gitclone.com/github.com/microsoft/autogen</a></td></tr>
|
||
<tr><td><strong>MetaGPT</strong></td><td>GitHub</td><td><a href="https://gitclone.com/github.com/geekan/MetaGPT" target="_blank">gitclone.com/github.com/geekan/MetaGPT</a></td></tr>
|
||
<tr><td><strong>LangGraph</strong></td><td>GitHub</td><td><a href="https://gitclone.com/github.com/langchain-ai/langgraph" target="_blank">gitclone.com/github.com/langchain-ai/langgraph</a></td></tr>
|
||
<tr><td><strong>AgentScope (阿里)</strong></td><td>GitHub</td><td><a href="https://gitclone.com/github.com/modelscope/agentscope" target="_blank">gitclone.com/github.com/modelscope/agentscope</a></td></tr>
|
||
<tr><td><strong>KYA</strong></td><td>PyPI</td><td><a href="https://pypi.org/project/veldt-kya" target="_blank">pypi.org/project/veldt-kya</a></td></tr>
|
||
<tr><td><strong>APBoa</strong></td><td>Gitee</td><td><a href="https://gitee.com/YuYanFeng/apboa" target="_blank">gitee.com/YuYanFeng/apboa</a></td></tr>
|
||
<tr><td><strong>Eigent</strong></td><td>GitHub</td><td><a href="https://gitclone.com/github.com/eigent-ai/eigent" target="_blank">gitclone.com/github.com/eigent-ai/eigent</a></td></tr>
|
||
</tbody>
|
||
</table>
|
||
</div>
|
||
|
||
<div class="table-wrap">
|
||
<div class="table-caption">表：低代码编排 & Harness基建项目</div>
|
||
<table>
|
||
<thead><tr><th style="width:14%">项目</th><th style="width:12%">领域</th><th>镜像地址</th></tr></thead>
|
||
<tbody>
|
||
<tr><td><strong>Open Agent Builder</strong></td><td>低代码</td><td><a href="https://gitclone.com/github.com/firecrawl/open-agent-builder" target="_blank">gitclone.com/github.com/firecrawl/open-agent-builder</a></td></tr>
|
||
<tr><td><strong>n8n</strong></td><td>低代码</td><td><a href="https://gitclone.com/github.com/n8n-io/n8n" target="_blank">gitclone.com/github.com/n8n-io/n8n</a></td></tr>
|
||
<tr><td><strong>Flowise</strong></td><td>低代码</td><td><a href="https://gitclone.com/github.com/FlowiseAI/Flowise" target="_blank">gitclone.com/github.com/FlowiseAI/Flowise</a></td></tr>
|
||
<tr><td><strong>RuoYi AI</strong></td><td>低代码</td><td><a href="https://gitee.com/zhddzrai/ruoyi-ai_1" target="_blank">gitee.com/zhddzrai/ruoyi-ai_1</a></td></tr>
|
||
<tr><td><strong>Conllect-LLM</strong></td><td>低代码</td><td><a href="https://gitee.com/one_love_op/conllect-llm" target="_blank">gitee.com/one_love_op/conllect-llm</a></td></tr>
|
||
<tr><td><strong>mem0</strong></td><td>记忆</td><td><a href="https://gitclone.com/github.com/mem0ai/mem0" target="_blank">gitclone.com/github.com/mem0ai/mem0</a></td></tr>
|
||
<tr><td><strong>Hindsight</strong></td><td>记忆</td><td><a href="https://gitclone.com/github.com/vectorize-io/hindsight" target="_blank">gitclone.com/github.com/vectorize-io/hindsight</a></td></tr>
|
||
<tr><td><strong>Letta (MemGPT)</strong></td><td>记忆</td><td><a href="https://gitclone.com/github.com/letta-ai/letta" target="_blank">gitclone.com/github.com/letta-ai/letta</a></td></tr>
|
||
<tr><td><strong>Zep / Graphiti</strong></td><td>记忆</td><td><a href="https://gitclone.com/github.com/getzep/graphiti" target="_blank">gitclone.com/github.com/getzep/graphiti</a></td></tr>
|
||
<tr><td><strong>GBrain</strong></td><td>记忆</td><td><a href="https://gitclone.com/github.com/garry-tan/gbrain" target="_blank">gitclone.com/github.com/garry-tan/gbrain</a></td></tr>
|
||
<tr><td><strong>ACE</strong></td><td>上下文</td><td><a href="https://gitclone.com/github.com/sambanova/ace" target="_blank">gitclone.com/github.com/sambanova/ace</a></td></tr>
|
||
<tr><td><strong>PRISM</strong></td><td>上下文</td><td><a href="https://gitclone.com/github.com/jingyip-cat/PRISM" target="_blank">gitclone.com/github.com/jingyip-cat/PRISM</a></td></tr>
|
||
<tr><td><strong>Google ADK</strong></td><td>会话</td><td><a href="https://gitclone.com/github.com/google/adk-python" target="_blank">gitclone.com/github.com/google/adk-python</a></td></tr>
|
||
<tr><td><strong>LangGraphChatBot</strong></td><td>会话</td><td><a href="https://gitee.com/NanGePlus/LangGraphChatBot" target="_blank">gitee.com/NanGePlus/LangGraphChatBot</a></td></tr>
|
||
<tr><td><strong>Agents-Flex</strong></td><td>Harness</td><td><a href="https://gitee.com/agents-flex/agents-flex" target="_blank">gitee.com/agents-flex/agents-flex</a></td></tr>
|
||
<tr><td><strong>ARES</strong></td><td>Harness</td><td><a href="https://gitee.com/dboy019/ares" target="_blank">gitee.com/dboy019/ares</a></td></tr>
|
||
</tbody>
|
||
</table>
|
||
</div>
|
||
|
||
<div class="table-wrap">
|
||
<div class="table-caption">表：技能市场 & 评测服务项目</div>
|
||
<table>
|
||
<thead><tr><th style="width:14%">项目</th><th style="width:12%">领域</th><th>镜像地址</th></tr></thead>
|
||
<tbody>
|
||
<tr><td><strong>SkillNet</strong></td><td>技能</td><td><a href="https://gitclone.com/github.com/zjunlp/SkillNet" target="_blank">gitclone.com/github.com/zjunlp/SkillNet</a></td></tr>
|
||
<tr><td><strong>skill-factory</strong></td><td>技能</td><td><a href="https://gitclone.com/github.com/rooftop-Owl/skill-factory" target="_blank">gitclone.com/github.com/rooftop-Owl/skill-factory</a></td></tr>
|
||
<tr><td><strong>awesome-agent-skills</strong></td><td>技能</td><td><a href="https://gitee.com/droidphone/awesome-agent-skills" target="_blank">gitee.com/droidphone/awesome-agent-skills</a></td></tr>
|
||
<tr><td><strong>witty-skill-insight</strong></td><td>技能</td><td><a href="https://gitcode.com/openeuler/witty-skill-insight" target="_blank">gitcode.com/openeuler/witty-skill-insight</a></td></tr>
|
||
<tr><td><strong>Future AGI</strong></td><td>评测</td><td><a href="https://gitclone.com/github.com/future-agi/future-agi" target="_blank">gitclone.com/github.com/future-agi/future-agi</a></td></tr>
|
||
<tr><td><strong>Opik (Comet)</strong></td><td>评测</td><td><a href="https://gitclone.com/github.com/comet-ml/opik" target="_blank">gitclone.com/github.com/comet-ml/opik</a></td></tr>
|
||
<tr><td><strong>EvalMonkey</strong></td><td>评测</td><td><a href="https://gitclone.com/github.com/Corbell-AI/evalmonkey" target="_blank">gitclone.com/github.com/Corbell-AI/evalmonkey</a></td></tr>
|
||
<tr><td><strong>Langfuse</strong></td><td>追踪</td><td><a href="https://gitclone.com/github.com/langfuse/langfuse" target="_blank">gitclone.com/github.com/langfuse/langfuse</a></td></tr>
|
||
<tr><td><strong>DeepEval</strong></td><td>评测</td><td><a href="https://gitclone.com/github.com/confident-ai/deepeval" target="_blank">gitclone.com/github.com/confident-ai/deepeval</a></td></tr>
|
||
<tr><td><strong>AgentBench (清华)</strong></td><td>评测</td><td><a href="https://gitclone.com/github.com/THUDM/AgentBench" target="_blank">gitclone.com/github.com/THUDM/AgentBench</a></td></tr>
|
||
</tbody>
|
||
</table>
|
||
</div>
|
||
|
||
<div class="box box-info">
|
||
<h4>ℹ️ 镜像说明</h4>
|
||
<p>GitHub 项目统一使用 <strong>gitclone.com</strong> 镜像（支持页面浏览 + Git 克隆）。如需克隆代码：</p>
|
||
<p style="font-family:'SF Mono',Monaco,monospace;font-size:12px;margin-top:8px;">
|
||
git clone https://gitclone.com/github.com/用户/仓库.git
|
||
</p>
|
||
<p style="margin-top:8px;">Gitee / GitCode 项目均为国内平台，使用原始地址即可直接访问。</p>
|
||
</div>
|
||
|
||
</section>
|
||
|
||
<footer class="footer">
|
||
<p>全功能智能体平台深度对比报告 · V1.0 · 2026-06-04</p>
|
||
<p style="margin-top:4px;">五大系统：多智能体平面 · 低代码编排 · Harness基建 · 技能市场 · 智能体评测 | 收录 81 个开源项目</p>
|
||
</footer>
|
||
|
||
</main>
|
||
|
||
<button class="back-to-top" id="backToTop" aria-label="回到顶部">↑</button>
|
||
|
||
<script>
|
||
(function(){
|
||
// Mobile drawer
|
||
const menuBtn = document.getElementById('menuBtn');
|
||
const overlay = document.getElementById('drawerOverlay');
|
||
const sidebar = document.querySelector('.sidebar');
|
||
let open = false;
|
||
function toggle(){ open ? close() : show(); }
|
||
function show(){ overlay.classList.add('show'); sidebar.classList.add('show'); open = true; document.body.style.overflow='hidden'; }
|
||
function close(){ overlay.classList.remove('show'); sidebar.classList.remove('show'); open = false; document.body.style.overflow=''; }
|
||
if(menuBtn) menuBtn.addEventListener('click', toggle);
|
||
if(overlay) overlay.addEventListener('click', close);
|
||
|
||
// Close drawer when a sidebar link is clicked (mobile)
|
||
const sidebarLinks = document.querySelectorAll('.sidebar nav a');
|
||
sidebarLinks.forEach(function(a){
|
||
a.addEventListener('click', function(){
|
||
if(window.innerWidth <= 768) close();
|
||
});
|
||
});
|
||
|
||
// Active TOC highlight on scroll
|
||
const sections = document.querySelectorAll('section[id]');
|
||
const navLinks = document.querySelectorAll('.sidebar nav a');
|
||
function onScroll(){
|
||
let current = '';
|
||
sections.forEach(function(sec){
|
||
if(window.scrollY >= sec.offsetTop - 100) current = sec.id;
|
||
});
|
||
navLinks.forEach(function(a){
|
||
a.classList.toggle('active', a.getAttribute('href') === '#' + current);
|
||
});
|
||
}
|
||
window.addEventListener('scroll', onScroll, {passive:true});
|
||
|
||
// Back to top
|
||
const btt = document.getElementById('backToTop');
|
||
function toggleBTT(){ btt.classList.toggle('show', window.scrollY > 400); }
|
||
window.addEventListener('scroll', toggleBTT, {passive:true});
|
||
if(btt) btt.addEventListener('click', function(){ window.scrollTo({top:0,behavior:'smooth'}); });
|
||
})();
|
||
</script>
|
||
</body>
|
||
</html> |