Add Qwen-Agent example to agent framework integrations (#4086)

* Add Qwen-Agent and lagent examples to agent framework integrations

This commit adds two new agent framework examples with Chinese language support:

1. **Qwen-Agent**: Alibaba Cloud's agent framework
   - Custom tool registration with @register_tool decorator
   - Calculator tool example demonstrating function calling
   - Support for OpenAI API and DashScope
   - MCP integration capabilities
   - Comprehensive Chinese and English documentation

2. **lagent**: InternLM's lightweight ReAct agent framework
   - ReAct reasoning pattern implementation
   - Built-in Python interpreter for code execution
   - Support for OpenAI and InternLM models
   - Lightweight design requiring minimal code
   - Full Chinese and English documentation

Both examples follow the established patterns:
- ZenML pipeline integration with ExternalArtifact
- Docker settings with UV package installer
- Error handling with success/error states
- Formatted response output
- Production-ready artifact management

The main README.md has been updated to include both frameworks in the
frameworks overview table and added a new "International Frameworks"
section highlighting Chinese language support.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <[email protected]>

* Update examples/agent_framework_integrations/README.md

Co-authored-by: Copilot <[email protected]>

* Apply suggestion from @strickvl

* Fix and test Qwen Agent

* deployers

* Delete lagent example

* clean up main README

* Linting fixes

---------

Co-authored-by: Claude <[email protected]>
Co-authored-by: Copilot <[email protected]>

Author: 3 people

examples/agent_framework_integrations/README.md

@@ -63,6 +63,7 @@ python run.py
 | [LlamaIndex](llama_index/) | 📚 Functions | Function agents, Async execution | llama-index, openai |
 | [OpenAI Agents SDK](openai_agents_sdk/) | 🏗️ Structured | Official OpenAI agents, Structured execution | openai-agents, openai |
 | [PydanticAI](pydanticai/) | ✅ Type-Safe | Type-safe agents, Validation | pydantic-ai, openai |
+| [Qwen-Agent](qwen-agent/) | 🧠 Function Call | Custom tools, MCP integration, Qwen models | qwen-agent, openai |
 | [Semantic Kernel](semantic-kernel/) | 🧩 Plugins | Plugin architecture, Microsoft ecosystem | semantic-kernel, openai |
 
 ## 🎯 Core Patterns

examples/agent_framework_integrations/qwen-agent/README.md

@@ -0,0 +1,113 @@
+# Qwen-Agent + ZenML
+
+## 中文快速开始 / Chinese Quick Start
+
+Qwen-Agent 是阿里巴巴开源的智能体框架，支持函数调用、工具使用和多模态能力。
+
+### 安装和运行
+```bash
+export OPENAI_API_KEY="your-api-key-here"
+cd qwen-agent/
+uv venv --python 3.11
+source .venv/bin/activate
+uv pip install -r requirements.txt
+```
+
+初始化 ZenML 并登录：
+```bash
+zenml init
+zenml login
+```
+
+运行示例流水线：
+```bash
+python run.py
+```
+
+### 部署流水线（实时服务）
+使用 ZenML 的部署功能将该流水线部署为 HTTP 实时服务：
+
+```bash
+# 将流水线部署为 HTTP 服务（引用 run.py 中的 agent_pipeline 符号）
+zenml pipeline deploy run.agent_pipeline --name qwen-agent
+```
+
+通过 CLI 调用服务：
+```bash
+zenml deployment invoke qwen-agent --query="计算 12*8 然后再加上 10"
+```
+
+通过 HTTP API 调用服务：
+```bash
+curl -X POST http://localhost:8000/invoke \
+  -H "Content-Type: application/json" \
+  -d '{"parameters": {"query": "请计算 15*7 并加上 42"}}'
+```
+
+> 提示：
+> - 该示例使用 OpenAI 兼容接口（在 `qwen_agent_impl.py` 中配置），需要设置环境变量 `OPENAI_API_KEY`。
+> - 如果改用阿里云 DashScope，可按照 Qwen-Agent 文档设置 `DASHSCOPE_API_KEY` 并相应调整 LLM 配置。
+
+---
+
+## English Documentation
+
+Qwen-Agent integrated with ZenML for production-grade agent orchestration.
+
+## Run
+```bash
+export OPENAI_API_KEY="your-api-key-here"
+uv venv --python 3.11
+source .venv/bin/activate
+uv pip install -r requirements.txt
+```
+
+Initialize ZenML and login:
+```bash
+zenml init
+zenml login
+```
+
+Run the pipeline locally:
+```bash
+python run.py
+```
+
+## Pipeline Deployment
+
+Deploy this agent as a real-time HTTP service:
+
+```bash
+# Deploy the pipeline as an HTTP service (referencing the agent_pipeline symbol in run.py)
+zenml pipeline deploy run.agent_pipeline --name qwen-agent
+```
+
+Invoke via CLI:
+```bash
+zenml deployment invoke qwen-agent --query="Calculate 25*4 and add 10"
+```
+
+Invoke via HTTP API:
+```bash
+curl -X POST http://localhost:8000/invoke \
+  -H "Content-Type: application/json" \
+  -d '{"parameters": {"query": "Add 42 to 7*15"}}'
+```
+
+## Features
+- Custom tool registration with `@register_tool` decorator
+- Function calling capabilities with Qwen models
+- Support for OpenAI-compatible APIs and DashScope
+- Built-in calculator tool for mathematical operations
+- Production-ready agent orchestration with ZenML
+
+## About Qwen-Agent
+
+Qwen-Agent is an open-source agent framework by Alibaba Cloud, built on the Qwen language models. It provides:
+- 🔧 Easy tool creation and registration
+- 🧠 Advanced function calling
+- 🌐 MCP (Model Context Protocol) integration
+- 📚 RAG (Retrieval-Augmented Generation) support
+- 💻 Code interpreter capabilities
+
+Learn more at the [official repository](https://github.com/QwenLM/Qwen-Agent).

examples/agent_framework_integrations/qwen-agent/qwen_agent_impl.py

@@ -0,0 +1,84 @@
+"""Qwen-Agent implementation with custom tools.
+
+This module demonstrates how to create a Qwen-Agent with custom tools
+for use in ZenML pipelines.
+"""
+
+import os
+
+import json5
+from qwen_agent.agents import Assistant
+from qwen_agent.tools.base import BaseTool, register_tool
+
+
+@register_tool("calculator")
+class Calculator(BaseTool):
+    """A simple calculator tool for basic arithmetic operations."""
+
+    name = "calculator"
+    description = "Performs basic arithmetic operations (add, subtract, multiply, divide)"
+    parameters = [
+        {
+            "name": "operation",
+            "type": "string",
+            "description": "The operation to perform: 'add', 'subtract', 'multiply', or 'divide'",
+            "required": True,
+        },
+        {
+            "name": "a",
+            "type": "number",
+            "description": "The first number",
+            "required": True,
+        },
+        {
+            "name": "b",
+            "type": "number",
+            "description": "The second number",
+            "required": True,
+        },
+    ]
+
+    def call(self, params: str, **kwargs) -> str:
+        # Qwen-Agent passes tool arguments as a JSON string; json5 allows mild deviations the LLM may produce.
+        data = json5.loads(params)
+        operation = data.get("operation")
+        a = float(data.get("a", 0))
+        b = float(data.get("b", 0))
+
+        if operation == "divide" and b == 0:
+            return "Error: Division by zero"
+
+        ops = {
+            "add": lambda x, y: x + y,
+            "subtract": lambda x, y: x - y,
+            "multiply": lambda x, y: x * y,
+            "divide": lambda x, y: x / y,
+        }
+        if operation not in ops:
+            return f"Error: Unknown operation '{operation}'"
+
+        result = ops[operation](a, b)
+        return str(result)
+
+
+# Initialize the LLM - uses OpenAI API by default
+llm_cfg = {
+    "model": "gpt-4o-mini",
+    # Explicit OpenAI base URL ensures consistent behavior across environments
+    "model_server": "https://api.openai.com/v1",
+    "api_key": os.environ.get("OPENAI_API_KEY"),
+    "generate_cfg": {
+        # You can add generation params here, e.g., "top_p": 0.8
+    },
+}
+
+system_instruction = (
+    "You are a helpful math assistant. Use the 'calculator' tool to perform arithmetic. "
+    "Respond with only the final numeric result unless the user requests detailed steps."
+)
+
+agent = Assistant(
+    llm=llm_cfg,
+    system_message=system_instruction,
+    function_list=["calculator"],
+)

examples/agent_framework_integrations/qwen-agent/requirements.txt

@@ -0,0 +1,4 @@
+qwen-agent>=0.0.10
+openai>=1.0.0
+zenml[local]
+json5>=0.9

examples/agent_framework_integrations/qwen-agent/run.py

@@ -0,0 +1,114 @@
+"""ZenML Pipeline for Qwen-Agent.
+
+This pipeline demonstrates how to integrate Qwen-Agent with ZenML
+for orchestration and artifact management.
+"""
+
+import os
+from typing import Annotated, Any, Dict
+
+from qwen_agent_impl import agent  # Use the local agent implementation
+
+from zenml import pipeline, step
+from zenml.config import DockerSettings, PythonPackageInstaller
+
+docker_settings = DockerSettings(
+    python_package_installer=PythonPackageInstaller.UV,
+    requirements="requirements.txt",  # relative to the pipeline directory
+    environment={
+        # Propagate API keys into the container so either OpenAI-compatible endpoints
+        # or Alibaba DashScope can be used interchangeably based on the LLM config.
+        "OPENAI_API_KEY": os.getenv("OPENAI_API_KEY"),
+        "DASHSCOPE_API_KEY": os.getenv("DASHSCOPE_API_KEY"),
+    },
+)
+
+
+@step
+def run_qwen_agent(
+    query: str,
+) -> Annotated[Dict[str, Any], "agent_results"]:
+    """Execute the Qwen-Agent with the given query."""
+    try:
+        messages = [{"role": "user", "content": query}]
+        last_batch: list[Any] = []
+        for batch in agent.run(messages=messages):
+            # run() yields incremental message lists; keep the latest for the final assistant output
+            last_batch = batch
+
+        final_text = ""
+        if last_batch:
+            last_msg = last_batch[-1]
+            final_text = (
+                last_msg.get("content", str(last_msg))
+                if isinstance(last_msg, dict)
+                else str(last_msg)
+            )
+
+        return {
+            "query": query,
+            "response": final_text,
+            "status": "success",
+        }
+    except Exception as e:
+        return {
+            "query": query,
+            "response": f"Agent error: {str(e)}",
+            "status": "error",
+        }
+
+
+@step
+def format_qwen_response(
+    agent_data: Dict[str, Any],
+) -> Annotated[str, "formatted_response"]:
+    """Format the Qwen-Agent results into a readable summary."""
+    query = agent_data["query"]
+    response = agent_data["response"]
+    status = agent_data["status"]
+
+    if status == "error":
+        formatted = f"""❌ QWEN-AGENT ERROR
+{"=" * 40}
+
+Query: {query}
+Error: {response}
+"""
+    else:
+        formatted = f"""🤖 QWEN-AGENT RESPONSE
+{"=" * 40}
+
+Query: {query}
+
+Response:
+{response}
+
+🧠 Powered by Qwen-Agent (Alibaba Cloud)
+"""
+
+    return formatted.strip()
+
+
+@pipeline(settings={"docker": docker_settings}, enable_cache=False)
+def agent_pipeline(
+    query: str = "Calculate the result of 15 multiplied by 7, then add 42 to it.",
+) -> str:
+    """ZenML pipeline that orchestrates the Qwen-Agent.
+
+    Returns:
+        Formatted agent response
+    """
+    # Run the Qwen-Agent with the provided query
+    agent_results = run_qwen_agent(query=query)
+
+    # Format the results
+    summary = format_qwen_response(agent_results)
+
+    return summary
+
+
+if __name__ == "__main__":
+    print("🚀 Running Qwen-Agent pipeline...")
+    run_result = agent_pipeline()
+    print("Pipeline completed successfully!")
+    print("Check the ZenML dashboard for detailed results and artifacts.")