Enables seamless cross-protocol AI agent communication by translating messages, intelligent routing, agent discovery, and secure authentication.
# Protocol Bridge SKILL / 协议桥技能
---
## English
### SKILL Overview
Protocol Bridge is an AI Agent communication infrastructure SKILL that enables seamless cross-protocol agent interaction.
### What This SKILL Does
This SKILL acts as a universal translator and router for AI agents, solving protocol incompatibility between MCP, A2A, LangChain, AutoGPT, and CrewAI.
### Key Capabilities
1. **Protocol Translation** - Convert messages between different protocol formats
2. **Intelligent Routing** - Route to agents based on capabilities
3. **Agent Discovery** - Automatic registration and capability matching
4. **Security** - JWT, OAuth, API keys, mTLS support
### Installation
```bash
npm install -g openclaw-protocol-bridge
```
### Basic Usage
#### Start Bridge Server
```bash
protocol-bridge serve --port 8080
```
#### Register Agents
```bash
# Register MCP agent
protocol-bridge register --id agent-1 --protocol mcp --endpoint http://localhost:3001
# Register A2A agent
protocol-bridge register --id agent-2 --protocol a2a --endpoint http://localhost:3002
```
#### Send Cross-Protocol Message
```typescript
import { ProtocolBridge } from 'openclaw-protocol-bridge';
const bridge = new ProtocolBridge({
protocols: {
mcp: { enabled: true },
a2a: { enabled: true }
}
});
await bridge.initialize();
const result = await bridge.send({
from: { id: 'agent-a', protocol: 'a2a' },
to: { id: 'agent-b', protocol: 'mcp' },
action: 'search',
params: { query: 'test' }
});
```
### Use Scenarios
**1. Enterprise Multi-Agent System**
Connect HR, Finance, and IT agents across different protocols.
**2. Multi-Framework Development**
Build agents with LangChain, AutoGPT, CrewAI - communicate seamlessly.
**3. Protocol Migration**
Migrate from one protocol to another without rewriting agents.
### Configuration
```json
{
"protocols": {
"mcp": { "enabled": true },
"a2a": { "enabled": true }
},
"routing": {
"strategy": "capability-based"
},
"security": {
"authentication": "jwt"
}
}
```
### CLI Commands
- `protocol-bridge serve` - Start bridge server
- `protocol-bridge init` - Initialize configuration
- `protocol-bridge register` - Register an agent
- `protocol-bridge list` - List all agents
- `protocol-bridge health` - Check system health
### Advanced Features
**Middleware Support**
```typescript
bridge.use(async (message, next) => {
console.log(`${message.from.id} → ${message.to.id}`);
return await next(message);
});
```
**Agent Discovery**
```typescript
const agents = await bridge.discover({
capabilities: ['search'],
protocols: ['mcp', 'a2a']
});
```
### Troubleshooting
**Agent Not Found**
```bash
protocol-bridge list
```
**Connection Timeout**
Increase timeout in config:
```json
{
"routing": {
"timeout": 60000
}
}
```
### Links
- GitHub: https://github.com/ZhenRobotics/openclaw-protocol-bridge
- npm: https://www.npmjs.com/package/openclaw-protocol-bridge
---
## 中文
### SKILL 概述
Protocol Bridge 是一个 AI Agent 通信基础设施 SKILL,可实现无缝的跨协议 Agent 交互。
### 此 SKILL 的功能
此 SKILL 充当 AI Agent 的通用翻译器和路由器,解决 MCP、A2A、LangChain、AutoGPT 和 CrewAI 之间的协议不兼容问题。
### 核心能力
1. **协议转换** - 在不同协议格式之间转换消息
2. **智能路由** - 根据 Agent 的能力进行路由
3. **Agent 发现** - 自动注册和能力匹配
4. **安全** - 支持 JWT、OAuth、API 密钥、mTLS
### 安装
```bash
npm install -g openclaw-protocol-bridge
```
### 基本使用
#### 启动桥接服务器
```bash
protocol-bridge serve --port 8080
```
#### 注册 Agent
```bash
# 注册 MCP Agent
protocol-bridge register --id agent-1 --protocol mcp --endpoint http://localhost:3001
# 注册 A2A Agent
protocol-bridge register --id agent-2 --protocol a2a --endpoint http://localhost:3002
```
#### 发送跨协议消息
```typescript
import { ProtocolBridge } from 'openclaw-protocol-bridge';
const bridge = new ProtocolBridge({
protocols: {
mcp: { enabled: true },
a2a: { enabled: true }
}
});
await bridge.initialize();
const result = await bridge.send({
from: { id: 'agent-a', protocol: 'a2a' },
to: { id: 'agent-b', protocol: 'mcp' },
action: 'search',
params: { query: 'test' }
});
```
### 使用场景
**1. 企业多 Agent 系统**
连接跨不同协议的人力资源、财务和 IT Agent。
**2. 多框架开发**
使用 LangChain、AutoGPT、CrewAI 构建 Agent - 无缝通信。
**3. 协议迁移**
从一个协议迁移到另一个协议,无需重写 Agent。
### 配置
```json
{
"protocols": {
"mcp": { "enabled": true },
"a2a": { "enabled": true }
},
"routing": {
"strategy": "capability-based"
},
"security": {
"authentication": "jwt"
}
}
```
### CLI 命令
- `protocol-bridge serve` - 启动桥接服务器
- `protocol-bridge init` - 初始化配置
- `protocol-bridge register` - 注册 Agent
- `protocol-bridge list` - 列出所有 Agent
- `protocol-bridge health` - 检查系统健康
### 高级功能
**中间件支持**
```typescript
bridge.use(async (message, next) => {
console.log(`${message.from.id} → ${message.to.id}`);
return await next(message);
});
```
**Agent 发现**
```typescript
const agents = await bridge.discover({
capabilities: ['search'],
protocols: ['mcp', 'a2a']
});
```
### 故障排查
**找不到 Agent**
```bash
protocol-bridge list
```
**连接超时**
在配置中增加超时:
```json
{
"routing": {
"timeout": 60000
}
}
```
### 链接
- GitHub: https://github.com/ZhenRobotics/openclaw-protocol-bridge
- npm: https://www.npmjs.com/package/openclaw-protocol-bridge
---
**v1.0.0** | MIT License | ZhenRobotics
don't have the plugin yet? install it then click "run inline in claude" again.