wow agent day08 初识zigent

Reference：Datawhale wow agent day08
Zigent 是一个基于 Agentlite 框架改进的智能代理开发框架。Agentlite 最初由 Salesforce AI Research 团队开发，是一个强大的 Agent 开发框架。Zigent 在其基础上进行了定制化改进，使其更适合特定场景的应用。
在本课中，我们将学习如何使用 Zigent 框架创建一个简单但功能完整的搜索代理。这个代理能够通过 DuckDuckGo 搜索引擎查找信息并回答问题。

环境准备#

import os
from dotenv import load_dotenv

# 加载环境变量
load_dotenv()
# 从环境变量中读取api_key
api_key = os.getenv('ZISHU_API_KEY')
base_url = "http://43.200.7.56:8008/v1"
chat_model = "glm-4-flash"

from typing import List
from zigent.agents import ABCAgent, BaseAgent
from zigent.llm.agent_llms import LLM
from zigent.commons import TaskPackage
from zigent.actions.BaseAction import BaseAction
from zigent.logging.multi_agent_log import AgentLogger
from duckduckgo_search import DDGS

配置 LLM#

这里使用 zigent 封装的 LLM 加载和配置 LLM 服务：

llm = LLM(api_key=api_key, base_url=base_url, model_name=chat_model)
response = llm.run("你是谁？")
print(response)

创建搜索动作#

class DuckSearchAction(BaseAction):
    def __init__(self) -> None:
        action_name = "DuckDuckGo_Search"
        action_desc = "Using this action to search online content."
        params_doc = {"query": "the search string. be simple."}
        self.ddgs = DDGS()
        super().__init__(
            action_name=action_name, 
            action_desc=action_desc, 
            params_doc=params_doc,
        )

    def __call__(self, query):
        results = self.ddgs.chat(query)
        return results

通过 call 方法执行实际的搜索操作使用示例：

search_action = DuckSearchAction()
results = search_action("什么是 agent")
print(results)

创建搜索代理#

我们创建一个继承自 BaseAgent 的搜索代理类，它需要一个大语言模型 (llm)、一组动作（默认是 DuckSearchAction）、代理名称和角色描述：

class DuckSearchAgent(BaseAgent):
    def __init__(
        self,
        llm: LLM,
        actions: List[BaseAction] = [DuckSearchAction()],
        manager: ABCAgent = None,
        **kwargs
    ):
        name = "duck_search_agent"
        role = "You can answer questions by using duck duck go search content."
        super().__init__(
            name=name,
            role=role,
            llm=llm,
            actions=actions,
            manager=manager
        )

执行代理#

def do_search_agent():
    # 创建代理实例
    search_agent = DuckSearchAgent(llm=llm)

    # 创建任务
    task = "what is the found date of microsoft"
    task_pack = TaskPackage(instruction=task)

    # 执行任务并获取响应
    response = search_agent(task_pack)
    print("response:", response)

if __name__ == "__main__":
    do_search_agent()

我们得到如下：

Agent [94mduck_search_agent[0m receives the following [4mTaskPackage[0m:
[96m[
	Task ID: e3a62788-ee7b-4495-9822-9530c5fdd799
	Instruction: what is the found date of microsoft
][0m
====[94mduck_search_agent[0m starts execution on TaskPackage e3a62788-ee7b-4495-9822-9530c5fdd799====
Agent [94mduck_search_agent[0m takes 0-step [4mAction[0m:
[94m{
	name: DuckDuckGo_Search
	params: {'query': 'Microsoft founding date'}
}[0m
Observation: [92mMicrosoft was founded on April 4, 1975.[0m
Agent [94mduck_search_agent[0m takes 1-step [4mAction[0m:
[94m{
	name: Finish
	params: {'response': 'Microsoft was founded on April 4, 1975.'}
}[0m
Observation: [92mTask Completed.[0m
=========[94mduck_search_agent[0m finish execution. TaskPackage[ID:e3a62788-ee7b-4495-9822-9530c5fdd799] status:
[96m[
	completion: completed
	answer: Microsoft was founded on April 4, 1975.
][0m
==========
response: Microsoft was founded on April 4, 1975.