Zai MCP Server D1e2419f

MCP运维云端

提供方 Zrald1 · 包名 zai-mcp-server@latest

运维云端 stdio · 本地进程 无需鉴权 社区 / 第三方 Claude
查看接入方式

ZAI多提供商MCP服务器是一个免费的多提供商AI服务器,支持OpenRouter、Anthropic和DeepSeek API,具有自动故障转移、AI到AI循环和智能数据收集功能。

一句话选型结论

按资料字段整理

Zai MCP Server D1e2419f 属于运维云端类 MCP 服务器,stdio · 本地进程,无需鉴权,可直接接入 Claude,有可直接执行的安装命令。接入前先确认权限范围;涉及写入、删除或付款的操作,建议先在测试环境验证。

分类运维云端 传输方式stdio · 本地进程 鉴权无需鉴权 部署方式本地部署
GitHub 星标 2 仓库热度,无公开仓库时显示未标注
工具数 未标注 暴露给模型的工具数量
资源 / 提示词 0 / 0 资源与提示词模板数量
维护方 社区 / 第三方 核验等级 仅资料核对

工具与能力

运维云端
数量来自来源目录的标注,具体工具清单以官方文档为准
工具数量未标注暴露给模型可调用的工具个数
资源数量0可读取的资源条目
提示词数量0内置提示词模板数量

能力标签

支持客户端

适用场景

AI服务器本地部署多提供商支持自动故障转移智能数据收集开源

安装与配置

stdio · 本地进程 无需鉴权
客户端在本机拉起子进程,数据不出本机
npx zai-mcp-server@latest

这条命令只是入口,具体启动参数请按官方文档补齐,不要直接照搬到别的服务器。

  1. 1
    准备运行环境

    stdio 服务在本地起进程,需要先装好 Node.js,再按安装命令拉取依赖。

  2. 2
    安装服务

    复制上面的安装命令执行;命令里带占位符的位置,换成自己的密钥或路径。

  3. 3
    写入客户端配置

    在客户端的 MCP 配置里填好启动命令和参数,保存后完全退出客户端再打开,让新配置生效。

  4. 4
    验证是否可用

    连接成功后随便调一个只读工具,确认能返回结果,再接入正式流程。

典型使用场景

3 类
按分类与标签整理的常见用法,具体以官方文档为准
云资源查询

查实例、存储、账单等云上资源状态。

部署与发布

把部署、重启、回滚这类操作接进对话。

监控与告警

拉取监控指标,异常时让模型帮你判断。

选择建议

选之前先看这几条
按条目字段给出的判断,不替你决定用不用

值得选的地方

  • 有现成的安装命令,不用自己找入口,照着跑就能起服务。
  • 已适配 Claude,换客户端不用重新研究配置。
  • 资料标注无需鉴权,装好就能连。

先想清楚的地方

  • stdio 方式要求本地装好运行时(如 Node.js),换机器要重新装一遍。
  • 资料没标注工具数量,实际能力要装好后确认。
  • 版本更新与行为细节以官方仓库为准。

更适合

  • 需要在运维云端场景里让模型直接调用外部能力的人。
  • 已经在用 Claude 的用户,接入路径最短。
  • 想先快速试一下、能接受在本地或自有环境跑服务的人。

不太适合

  • 完全不想在本地装环境的人。
  • 需要厂商级 SLA 和技术支持的生产环境。

同类对比

6 个同类
同分类里挑最接近的几条,字段以各自详情页为准
MCP 服务器工具数传输方式部署方式安装命令最后核验
Zai MCP Server D1e2419f当前 未标注 stdio · 本地进程 本地部署 有 2026-05-17
lobehub 未标注 stdio · 本地进程 本地部署 有 2026-04-11
TrendRadar 未标注 stdio · 本地进程 本地部署 有 2026-04-13
Ui Tars Desktop 8a825770 未标注 stdio · 本地进程 本地部署 有 2026-05-17
ruflo 未标注 stdio · 本地进程 本地部署 有 2026-04-13
FastGPT 未标注 stdio · 本地进程 本地部署 有 2026-04-11

服务器档案

运维云端
字段来自项目仓库与公开索引,中文说明由本站标注
提供方Zrald1上游标注的维护方 Zrald1
分类运维云端本站按上游分类展示
连接方式stdio · 本地进程客户端在本机拉起子进程,数据不出本机
鉴权方式无需鉴权远程服务建议用最小权限凭据
维护方社区 / 第三方核验等级 仅资料核对
包名 / 运行时zai-mcp-server@latest · Node.js本地安装时用到的包标识与运行时
部署类型本地部署上游给出的部署形态
最近核验2026-05-17上游最后核验时间

详细介绍

16.3 KB
来自项目仓库与上游目录的原文,未做删改

🤖 ZAI Multi-Provider MCP Server

FREE Multi-Provider AI MCP Server with support for OpenRouter, Anthropic, and DeepSeek APIs. Features automatic failover, AI-to-AI loops, and smart data collection. No license validation required - completely free for all users!

💝 Support Development

If you find this project helpful, consider supporting development:

Crypto Donations

  • BNB (Binance Smart Chain): 0xB8E0b6D4BaaFd1ac4De9245A760cB8F09bB7D084
  • Bitcoin (BTC): bc1q77k0ju6ta3sp0vm3phm6dek432rzg7cqwf43z6
  • Ethereum (ETH): 0xB8E0b6D4BaaFd1ac4De9245A760cB8F09bB7D084
  • Polygon (MATIC): 0xB8E0b6D4BaaFd1ac4De9245A760cB8F09bB7D084
  • Dogecoin (DOGE): DGbDQEgJLnNR2yEmWFrusFd7jLya4aoZMA

Your support helps maintain and improve this open-source project! 🙏

✨ Key Features

  • 🆓 Completely FREE - No license validation or restrictions
  • 🤖 Multi-Provider Support - OpenRouter, Anthropic, DeepSeek APIs
  • 🔄 Automatic Failover - Smart switching between providers/models
  • 🔁 AI-to-AI Loops - Infinite improvement cycles
  • 📊 Smart Data Collection - Contributes to AI model development
  • ⚡ High Availability - Multiple API keys with rotation
  • 🎯 Quality Filtering - Only valuable interactions collected
  • 🌐 Global Access - Works worldwide, no restrictions

🚀 Quick Setup

Option 1: OpenRouter (Recommended - Free Models Available)```json

{
"mcpServers": {
"zai-mcp-server": {
"command": "npx",
"args": ["-y", "zai-mcp-server@latest"],
"env": {
"OPENROUTER_API_KEY": "sk-or-v1-abc123...,sk-or-v1-def456...,sk-or-v1-ghi789...",
"MODEL": "deepseek/deepseek-r1-0528:free"
}
}
}
}


### **Option 2: Anthropic Claude**```json
{
  "mcpServers": {
    "zai-mcp-server": {
      "command": "npx",
      "args": ["-y", "zai-mcp-server@latest"],
      "env": {
        "ANTHROPIC_API_KEY": "sk-ant-api03-abc123...",
        "MODEL": "claude-3-5-sonnet-20241022"
      }
    }
  }
}

Option 3: Google Gemini (Advanced Multimodal)```json

{
"mcpServers": {
"zai-mcp-server": {
"command": "npx",
"args": ["-y", "zai-mcp-server@latest"],
"env": {
"GEMINI_API_KEY": "AIzaSyAbc123...",
"MODEL": "gemini-2.5-flash-preview-05-20"
}
}
}
}


### **Option 4: DeepSeek (Best Value)**```json
{
  "mcpServers": {
    "zai-mcp-server": {
      "command": "npx",
      "args": ["-y", "zai-mcp-server@latest"],
      "env": {
        "DEEPSEEK_API_KEY": "sk-abc123...",
        "MODEL": "deepseek-chat"
      }
    }
  }
}

Option 5: Multi-Model Voting (AI Consensus)```json

{
"mcpServers": {
"zai-mcp-server": {
"command": "npx",
"args": ["-y", "zai-mcp-server@latest"],
"env": {
"OPENROUTER_API_KEY": "sk-or-v1-abc123...,sk-or-v1-def456...,sk-or-v1-ghi789...",
"ANTHROPIC_API_KEY": "sk-ant-api03-abc123...",
"DEEPSEEK_API_KEY": "sk-abc123...",
"MODEL": "voting-consensus",
"VOTING_PANEL": "general",
"VOTING_STRATEGY": "consensus"
}
}
}
}


## 📋 **Supported Models**

### **OpenRouter Models (2025 Updated)**

#### **🆓 FREE Models (Recommended)**
- `deepseek/deepseek-r1-0528:free` - **NEW** Latest DeepSeek R1 reasoning model (671B params)
- `deepseek/deepseek-r1-0528-qwen3-8b:free` - **NEW** Distilled 8B reasoning model
- `mistralai/devstral-small:free` - **NEW** 24B coding-focused model (SWE-Bench optimized)
- `sarvamai/sarvam-m:free` - **NEW** Multilingual model with reasoning (24B params)
- `google/gemma-3n-e4b-it:free` - **NEW** Google's latest Gemma model
- `meta-llama/llama-3.3-8b-instruct:free` - **UPDATED** Meta's latest Llama
- `microsoft/phi-4-reasoning:free` - **NEW** Microsoft's reasoning model
- `microsoft/phi-4-reasoning-plus:free` - **NEW** Enhanced reasoning model
- `qwen/qwen3-8b:free` - **NEW** Qwen3 8B model
- `qwen/qwen3-14b:free` - **NEW** Qwen3 14B model
- `qwen/qwen3-30b-a3b:free` - **NEW** Qwen3 30B model
- `qwen/qwen3-32b:free` - **NEW** Qwen3 32B model
- `thudm/glm-z1-32b:free` - **NEW** GLM reasoning model (32B params)

#### **💎 Premium Models**
- `anthropic/claude-opus-4` - **NEW** Most powerful Claude model (2025)
- `anthropic/claude-sonnet-4` - **NEW** High performance Claude (2025)
- `google/gemini-2.5-pro-preview` - **NEW** Google's latest Gemini Pro
- `google/gemini-2.5-flash-preview-05-20` - **NEW** Fast Gemini model
- `openai/gpt-4o` - OpenAI's flagship model
- `openai/gpt-4o-mini` - Compact and powerful
- `openai/o1-preview` - **NEW** OpenAI's reasoning model
- `openai/o1-mini` - **NEW** Compact reasoning model
- `anthropic/claude-3-5-sonnet-20241022` - Recommended balance
- `anthropic/claude-3-5-haiku-20241022` - Fastest Claude
- `deepseek/deepseek-chat` - General purpose (DeepSeek-V3)
- `deepseek/deepseek-reasoner` - **NEW** Advanced reasoning (DeepSeek-R1)

### **Anthropic Models (2025)**
- `claude-opus-4-20250514` - Most powerful (newest)
- `claude-sonnet-4-20250514` - High performance (newest)
- `claude-3-5-sonnet-20241022` - Recommended balance
- `claude-3-5-haiku-20241022` - Fastest and cheapest

**Note**: Anthropic models are also available through OpenRouter with the following IDs:
- `anthropic/claude-opus-4` - Claude Opus 4 via OpenRouter
- `anthropic/claude-sonnet-4` - Claude Sonnet 4 via OpenRouter
- `anthropic/claude-3-5-sonnet-20241022` - Claude 3.5 Sonnet via OpenRouter
- `anthropic/claude-3-5-haiku-20241022` - Claude 3.5 Haiku via OpenRouter

### **DeepSeek Models (2025)**
- `deepseek-chat` - General purpose (DeepSeek-V3)
- `deepseek-reasoner` - Advanced reasoning (DeepSeek-R1)

**Note**: DeepSeek models are also available FREE through OpenRouter:
- `deepseek/deepseek-r1-0528:free` - **FREE** Latest R1 reasoning model (671B params)
- `deepseek/deepseek-r1-0528-qwen3-8b:free` - **FREE** Distilled 8B version
- `deepseek/deepseek-chat` - General purpose via OpenRouter (paid)
- `deepseek/deepseek-reasoner` - Advanced reasoning via OpenRouter (paid)

### **Google Gemini Models (2025 - Confirmed Working)**
- `gemini-2.5-flash-preview-05-20` - **LATEST** Most advanced multimodal model (May 2025)
- `gemini-2.0-flash` - **STABLE** Next-gen features with enhanced speed
- `gemini-2.0-flash-001` - **STABLE** Versioned 2.0 Flash model
- `gemini-2.0-flash-lite` - **FAST** Optimized for speed and cost efficiency
- `gemini-1.5-flash-latest` - **RELIABLE** Fast and versatile multimodal
- `gemini-1.5-flash-8b-latest` - **LIGHTWEIGHT** Efficient 8B parameter model
- `gemini-1.5-flash` - **PRODUCTION** Stable Flash model
- `gemini-1.5-flash-8b` - **EFFICIENT** Stable 8B model

**Experimental Models** (May have rate limits):
- `gemini-2.5-pro-preview-06-05` - **PREVIEW** Most powerful reasoning model
- `gemini-2.0-flash-thinking-exp-01-21` - **EXPERIMENTAL** Advanced thinking model
- `gemini-2.0-flash-exp` - **EXPERIMENTAL** Latest experimental features

**Note**: All models tested and confirmed working with the provided API key. Gemini models support multimodal inputs (text, images, audio, video) and provide excellent performance for various tasks.

## 🆕 **What's New in 2025**

### **🚀 Latest Model Updates**
- **13 FREE models** now available through OpenRouter
- **4 CONFIRMED Gemini models** tested and working (2025)
- **DeepSeek R1 0528**: Latest reasoning model with 671B parameters
- **Gemini 2.5 Flash Preview**: Google's most advanced multimodal model
- **Mistral Devstral Small**: 24B coding-focused model optimized for SWE-Bench
- **Microsoft Phi-4**: New reasoning models with enhanced capabilities
- **Qwen3 Series**: Multiple variants (8B, 14B, 30B, 32B) all available for free
- **Anthropic Claude 4**: Opus and Sonnet variants now available
- **Google Gemini 2.0 Flash**: Next-generation features with enhanced speed

### **🔧 Enhanced Features**
- **🗳️ Multi-Model AI Voting**: Multiple AI models vote on best responses
- **🤖 AI Agent Panels**: Specialized agent groups (coding, reasoning, general, gemini)
- **📊 Consensus Algorithms**: Multiple voting strategies (majority, consensus, weighted)
- **🎯 Smart Agent Selection**: Performance-based agent selection
- **🌐 Multi-Provider Support**: OpenRouter + Google Gemini + DeepSeek + Anthropic
- **Automatic Model Failover**: Seamlessly switches between providers
- **Smart API Key Rotation**: Supports multiple keys per provider
- **Enhanced Error Handling**: Better recovery from API failures
- **Real-time Status Monitoring**: Track provider health and usage
- **Improved Data Collection**: Better quality filtering for AI training

## 🛠️ **Installation**

### **VSCode MCP Configuration**

1. **Open VSCode Settings** (Ctrl/Cmd + ,)
2. **Search for "MCP"** or go to Extensions → MCP
3. **Add the configuration** above to your MCP settings
4. **Restart VSCode** to activate

### **Alternative: Direct Installation**```bash
# Install globally
npm install -g zai-mcp-server

# Or run directly
npx zai-mcp-server@latest

🎯 Available Tools

🔄 AI-to-AI Loop Tools

  • activate_infinite_loop - Start AI-to-AI improvement loops
  • stop_ai_loops - Stop all active loops
  • list_active_loops - View running loops
  • get_ai_prompts - Get AI-generated prompts
  • acknowledge_agent_response - Process AI responses

🗳️ Multi-Model Voting Tools

  • ai_voting_request - Submit prompt for multi-model AI consensus
  • get_voting_history - View recent voting sessions
  • get_agent_performance - Check AI agent performance stats

🤖 Provider Management Tools

  • get_ai_provider_status - Check provider status
  • reset_ai_providers - Reset failed providers

🗳️ AI Voting System

How It Works

The ZAI MCP Server features an advanced multi-model voting system where multiple AI agents collaborate to provide the best possible responses:

  1. 🤖 Agent Selection: System selects specialized AI agents based on the task
  2. 📝 Response Generation: Each agent generates their own response
  3. 🗳️ Voting Phase: All agents vote on which response is best
  4. 🎯 Consensus: System calculates consensus and selects winner
  5. 📊 Learning: Agent performance is tracked and improved over time

Voting Panels

🔧 General Panel (Default)

  • DeepSeek R1: Reasoning specialist (671B params)
  • Mistral Devstral: Coding expert (24B params)
  • Microsoft Phi-4: Analysis agent
  • Qwen3 14B: General assistant
  • Llama 3.3: Conversation expert

💻 Coding Panel

  • Mistral Devstral: Lead coding specialist
  • DeepSeek R1 Qwen: Reasoning + coding
  • Microsoft Phi-4+: Development expert
  • Qwen3 32B: Large model for complex projects

🧠 Reasoning Panel

  • DeepSeek R1: State-of-the-art reasoning
  • Microsoft Phi-4: Analytical thinking
  • GLM Z1: Creative reasoning
  • Qwen3 30B: Data analysis

💎 Premium Panel (Requires Paid APIs)

  • Claude Opus 4: Most powerful model
  • OpenAI o1: Advanced reasoning
  • Gemini 2.5 Pro: Multimodal capabilities
  • Claude Sonnet 4: High performance

Voting Strategies

  • 🗳️ Majority: Simple majority wins (50%+ threshold)
  • 🤝 Consensus: Strong agreement required (70%+ threshold)
  • ⚖️ Weighted: Votes weighted by agent expertise and confidence
  • 💯 Unanimous: All agents must agree (100% threshold)

Example Usage```javascript

// Request AI voting on a coding problem
await ai_voting_request({
prompt: "Optimize this React component for performance",
panel: "coding",
strategy: "consensus",
maxAgents: 4
});

// Get voting history
await get_voting_history({ limit: 5 });

// Check agent performance
await get_agent_performance();


## 💡 **Usage Examples**

### **Start AI-to-AI Loop**```
Use the "activate_infinite_loop" tool with:
- message: "actloop improve my React component performance"
- aiToAi: true

Check Provider Status```

Use the "get_ai_provider_status" tool to see:
- Current provider and model
- Available API keys
- Failed providers
- Request statistics


### **Stop Loops**```
Use "stop_ai_loops" with:
- message: "stploop"

📊 AI Model Development Contribution

Important Notice: By using the ZAI MCP Server, you acknowledge and agree that AI-to-AI interactions facilitated by this server may be utilized for AI model development, research, and improvement initiatives. This contributes to the advancement of artificial intelligence technology and helps create better AI systems for the community.

What Gets Collected:

  • ✅ AI-to-AI problem-solving conversations
  • ✅ Code generation and improvement examples
  • ✅ Multi-iteration debugging sessions
  • ✅ High-quality interactions (80%+ score)

What Gets Filtered Out:

  • ❌ Low-quality responses
  • ❌ Error-heavy conversations
  • ❌ Personal information
  • ❌ Non-problem-solving interactions

Data Usage:

  • Training data is used to improve AI models
  • Helps advance AI-to-AI collaboration research
  • Contributes to open AI development
  • All usage complies with applicable data protection regulations

🔧 Configuration Options

Environment Variables

  • OPENROUTER_API_KEY - Comma-separated OpenRouter keys
  • ANTHROPIC_API_KEY - Comma-separated Anthropic keys
  • DEEPSEEK_API_KEY - Comma-separated DeepSeek keys
  • MODEL - Primary model to use
  • ZAI_FREE_MODE - Always true (no license needed)
  • ZAI_DATA_COLLECTION - Always true (automatic)

Multiple API Keys```json

"OPENROUTER_API_KEY": "key1,key2,key3,key4"

The server automatically rotates between keys for high availability.

## 🚀 **Advanced Features**

### **Automatic Failover**
- Switches between providers when one fails
- Rotates API keys automatically
- Tries different models for best results

### **Smart Data Collection**
- Only collects valuable AI interactions
- Filters out errors and low-quality responses
- Compresses and stores efficiently

### **High Availability**
- Multiple API providers
- Multiple keys per provider
- Automatic error recovery

## 🆓 **Why It's Free**

This MCP server is completely free because:
- **No License Validation** - No restrictions or paywalls
- **Community Driven** - Open source development
- **Data Collection** - Valuable training data helps fund development
- **AI Advancement** - Contributes to AI research and development

## 📊 **Data Collection**

This server automatically collects valuable AI-to-AI interactions for training data:

### **What Gets Collected:**
- ✅ AI-to-AI problem-solving conversations
- ✅ Code generation and improvement examples
- ✅ Multi-iteration debugging sessions
- ✅ High-quality interactions (80%+ score)

### **What Gets Filtered Out:**
- ❌ Low-quality responses
- ❌ Error-heavy conversations
- ❌ Personal information
- ❌ Non-problem-solving interactions

### **Data Usage:**
- Training data is used to improve AI models
- Helps advance AI-to-AI collaboration research
- Contributes to open AI development

## 🔧 **Configuration Options**

### **Environment Variables**
- `OPENROUTER_API_KEY` - Comma-separated OpenRouter keys
- `ANTHROPIC_API_KEY` - Comma-separated Anthropic keys
- `DEEPSEEK_API_KEY` - Comma-separated DeepSeek keys
- `MODEL` - Primary model to use
- `ZAI_FREE_MODE` - Always true (no license needed)
- `ZAI_DATA_COLLECTION` - Always true (automatic)

### **Multiple API Keys**```json
"OPENROUTER_API_KEY": "key1,key2,key3,key4"

The server automatically rotates between keys for high availability.

🚀 Advanced Features

Automatic Failover

  • Switches between providers when one fails
  • Rotates API keys automatically
  • Tries different models for best results

Smart Data Collection

  • Only collects valuable AI interactions
  • Filters out errors and low-quality responses
  • Compresses and stores efficiently

High Availability

  • Multiple API providers
  • Multiple keys per provider
  • Automatic error recovery

🆓 Why It's Free

This MCP server is completely free because:
- No License Validation - No restrictions or paywalls
- Community Driven - Open source development
- Data Collection - Valuable training data helps fund development
- AI Advancement - Contributes to AI research and development

🤝 Contributing

We welcome contributions! This project helps advance AI-to-AI collaboration research.

📄 License

MIT License - Use freely in any project, commercial or personal.

🔗 Links

  • NPM: zai-mcp-server

🎉 Start using your FREE multi-provider AI MCP server today!

常见问题

8 问
答案由本站依据条目字段整理,不引用上游原文
Zai MCP Server D1e2419f 怎么安装?

在客户端按下面的安装命令接入:npx zai-mcp-server@latest。对应的包名是 zai-mcp-server@latest。装完在客户端配置里引用同一个包或端点。

Zai MCP Server D1e2419f 怎么在 Claude 里配置?

它的传输方式是stdio · 本地进程,在客户端的 MCP 配置里填好启动命令和参数,保存后完全退出客户端再打开,配置才会生效。

Zai MCP Server D1e2419f 收费吗?

本站不代收费用,也没有标注具体价格。是否收费取决于服务本身和背后的接口额度,正式使用前按官方说明确认。

Zai MCP Server D1e2419f 需要 API Key 吗?

资料标注为无需认证,装好就能连;如果实际调用时报鉴权错误,按官方文档补上凭据。

Zai MCP Server D1e2419f 支持哪些客户端?

资料标注支持 Claude;其他客户端只要能配置 MCP Server,也能用同样的命令或地址接入。

Zai MCP Server D1e2419f 能提供哪些能力?

资料显示能力标签包括 JavaScript、Claude、开源。具体工具清单以官方文档为准。

Zai MCP Server D1e2419f 需要什么运行环境?

运行时为 Node.js;属于本地进程,需要在本机装好运行环境。

Zai MCP Server D1e2419f 是本地还是远程?

它是本地进程(stdio),由客户端在本机拉起。传输方式为stdio · 本地进程。

热门专题

按标签挑服务器
全部专题 →

服务器信息以作者仓库与官方文档为准;分类、传输方式与鉴权方式的中文说明为中文整理,最近核验 2026-05-17。