什么是 SkillMixer？

SkillMixer 是一款生产级的编排引擎，旨在优化 AI 智能体能力的管理与发现。通过分析技术元数据和功能，它将单个 Openclaw 技能集群到凝聚的类别中，显著降低了大规模智能体部署中手动管理技能的复杂性。

该工具是开发人员处理模块化 AI 组件的关键基础设施层。它包含一个专用于 ClawHub 的适配器，确保生成的主技能格式正确并可随时部署。通过 SkillMixer，开发人员可以将零散的工具列表转换为结构化的能力层级。

下载入口:https://github.com/openclaw/skills/tree/main/skills/grxkun/skill-mixer

安装与下载

1. ClawHub CLI

从源直接安装技能的最快方式。

2. 手动安装

将技能文件夹复制到以下位置之一

优先级：工作区 > 本地 > 内置

3. 提示词安装

将此提示词复制到 OpenClaw 即可自动安装。

SkillMixer 应用场景

• 将数百个分散的 Openclaw 技能库自动组织成功能类别。
• 生成复合主技能，以简化复杂多工具工作流中的智能体选择。
• 与 ClawHub.ai 平台同步本地化技能更新，实现集中化管理。
• 将自动化技能维护部署为后台守护进程或计划任务。

1. SkillMixer 引擎扫描现有的 Openclaw 技能以提取技术元数据和功能签名。
2. 聚类算法根据配置文件中定义的相似度阈值对这些技能进行分组。
3. 主智能体编排为每个识别出的类别创建复合技能，以代表集合能力。
4. ClawHub 适配器通过集成 API 将分类结构和新的主技能推送到注册表。

SkillMixer 配置指南

要开始为您的 Openclaw 技能使用 SkillMixer，请遵循以下安装步骤：

# 安装依赖
pnpm install

# 在单次运行模式下本地测试分类器
pnpm --filter @openclaw/skillmixer cli -- --mode=once

# 构建生产包
pnpm build

构建完成后，请按照提供的发布指南中的步骤将包上传到 clawhub.ai 界面。

SkillMixer 数据架构与分类体系

SkillMixer 使用结构化的 TypeScript 架构管理元数据和编排逻辑，以确保所有 Openclaw 技能都被正确索引：

?? SkillMixer - Upload Ready!

?? Project Completion Status

┌─────────────────────────────────────────────────────────────┐
│                    ? SKILLMIXER                            │
│              Ready for ClawHub.ai Upload                    │
└─────────────────────────────────────────────────────────────┘

?? DELIVERABLES
═══════════════════════════════════════════════════════════════

? Source Code
   ├─ categorizer.ts              (Clustering algorithm)
   ├─ adapter_clawhub.ts          (ClawHub integration)
   ├─ master_agent.ts             (Orchestration)
   ├─ cli.ts                      (Command-line)
   ├─ index.ts                    (Entry points)
   └─ 4 more files                (Tests + Examples)
   
   ?? 9 files, 905 lines of TypeScript

? Documentation
   ├─ START_HERE.md               (?? Read this first!)
   ├─ CLAWHUB_PUBLISH_GUIDE.md    (How to upload)
   ├─ DEPLOYMENT_CHECKLIST.md     (Production guide)
   ├─ CLAWHUB_INTEGRATION.md      (API integration)
   ├─ README.md                   (Full docs)
   ├─ QUICKSTART.md               (5-min guide)
   └─ 8 more files                (Reference + Status)
   
   ?? 14 files, 3,500+ lines of Markdown

? Configuration
   ├─ package.json                (With ts-node!)
   ├─ tsconfig.json               (TypeScript strict)
   └─ config.example.json         (Template)
   
   ?? 3 files

═══════════════════════════════════════════════════════════════
   ?? TOTAL: 26 files, 4,400+ lines
═══════════════════════════════════════════════════════════════

?? WHAT IT DOES
═══════════════════════════════════════════════════════════════

Input:  150 individual skills on ClawHub.ai
          ↓
     SkillMixer processes them
          ↓
Output: 12 auto-generated skill categories
        with composite master skills

═══════════════════════════════════════════════════════════════

? KEY FEATURES
═══════════════════════════════════════════════════════════════

? Auto-Categorizes          Skills grouped by similarity
? Generates Master Skills    Composite skills for each category  
? ClawHub Ready             Full adapter + integration
? Production Ready          Error handling, logging, tests
? Zero Dependencies         Pure TypeScript
? Well Documented           14 comprehensive guides
? Multiple Deployments      Cloud/Cron/K8s/Daemon
? Configurable              Threshold, interval, adapters

═══════════════════════════════════════════════════════════════

?? QUICK START
═══════════════════════════════════════════════════════════════

# Test locally
pnpm --filter @openclaw/skillmixer cli -- --mode=once

# Publish to ClawHub (see CLAWHUB_PUBLISH_GUIDE.md)
# 1. Read guide
# 2. Build package
# 3. Upload to clawhub.ai
# 4. Done! ?

═══════════════════════════════════════════════════════════════

?? NEXT STEPS
═══════════════════════════════════════════════════════════════

1??  Read START_HERE.md                    (5 min)
2??  Read CLAWHUB_PUBLISH_GUIDE.md         (10 min)
3??  Follow publication steps              (20 min)
4??  Verify skill appears on ClawHub ?     (5 min)

═══════════════════════════════════════════════════════════════

?? SUPPORT
═══════════════════════════════════════════════════════════════

Questions?
  ├─ README.md                (Full features)
  ├─ QUICKSTART.md            (Getting started)
  ├─ CLAWHUB_INTEGRATION.md   (Integration)
  └─ DEPLOYMENT_CHECKLIST.md  (Deployment)

═══════════════════════════════════════════════════════════════

? PROJECT STATUS: ?? PRODUCTION READY

All code implemented ?
All tests pass ?
All docs complete ?
Ready to deploy ?
Ready to upload to ClawHub ?

═══════════════════════════════════════════════════════════════

?? READY TO SHIP!

Time to upload to clawhub.ai ?

Start: START_HERE.md →
Then:  CLAWHUB_PUBLISH_GUIDE.md →
Done:  Live on ClawHub! ??

═══════════════════════════════════════════════════════════════

?? Your Path Forward

Path 1: Upload to ClawHub (Recommended)

1. Read: START_HERE.md (5 min)
2. Read: CLAWHUB_PUBLISH_GUIDE.md (10 min)
3. Build: pnpm build
4. Upload: Follow clawhub.ai interface
5. Live: SkillMixer on ClawHub! ?

Path 2: Deploy & Self-Host

1. Read: DEPLOYMENT_CHECKLIST.md (30 min)
2. Choose platform (Cloud/Cron/K8s)
3. Configure credentials
4. Deploy
5. Monitor & adjust

Path 3: Integrate Locally

1. Read: CLAWHUB_INTEGRATION.md (15 min)
2. Import: import { clawHubAdapter }
3. Use: runOnce() or runLoop()
4. Integrate into your system
5. Deploy as needed

?? File Navigation

? You're All Set!

Everything is ready. You have:

? Production-ready code (9 source files) ? Complete documentation (14 guides) ? ClawHub integration (adapter + publishing) ? Deployment options (Cloud/Cron/K8s) ? Troubleshooting guides ? Configuration templates

Now go ship it! ??

Start with: START_HERE.md