搭建农村电商平台免费seo营销软件

实战简单的agent示例

代码功能概述

• 目标：创建一个能自主思考（推理）并调用工具（行动）的 AI Agent，用于回答需要实时信息的问题（如“Agent 最新研究进展”）。

• 核心组件：

  • LLM（GPT-4）：负责推理和决策。

  • SerpAPI（搜索引擎）：当 LLM 缺乏知识时，用于搜索外部信息。

  • ReAct 框架：让 Agent 循环执行“思考→行动→观察→优化”。

from dotenv import load_dotenv
load_dotenv() #load_dotenv() 加载 .env 文件中的环境变量（如 OpenAI 和 SerpAPI 的 API 密钥）from langchain import hub
from langchain_openai import ChatOpenAI
from langchain_community.utilities import SerpAPIWrapper
from langchain_core.tools import Tool
from langchain.agents import create_react_agent, AgentExecutor# Initialize prompt
prompt = hub.pull("hwchase17/react")
print(prompt)   # 从 LangChain Hub 加载预定义的 ReAct 提示模板（hwchase17/react），该模板包含思考、工具调用和最终回答的引导语。# Initialize LLM - using ChatOpenAI instead of OpenAI for better compatibility
llm = ChatOpenAI(model="gpt-4o-mini",  # Use a known working modeltemperature=0
)# Initialize search tool
search = SerpAPIWrapper()
tools = [Tool(name="Search",func=search.run,description="当大模型没有相关知识时，用于搜索知识"),
]# Create agent
agent = create_react_agent(llm, tools, prompt)
agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True)#通过 AgentExecutor 执行两次相同查询，观察结果一致性。
#verbose=True 会打印工具调用和 LLM 输出的详细日志，便于调试。# Run queries
print("第一次运行的结果：")
try:result1 = agent_executor.invoke({"input": "当前Agent最新研究进展是什么?"})print(result1)
except Exception as e:print(f"Error in first query: {e}")print("第二次运行的结果：")
try:result2 = agent_executor.invoke({"input": "当前Agent最新研究进展是什么？"})print(result2)
except Exception as e:print(f"Error in second query: {e}")

 自行解决各类api key

.env 文件

OPENAI_API_KEY="sk-xxx"
OPENAI_API_BASE="https:xxx"
SERPAPI_API_KEY="1abd77afxxxxx"

有记忆版本：

import sys
from dotenv import load_dotenv
load_dotenv()from langchain.prompts import PromptTemplate
from langchain_openai import ChatOpenAI
from langchain_community.utilities import SerpAPIWrapper
from langchain_core.tools import Tool
from langchain.agents import create_react_agent, AgentExecutor
from langchain.memory import ConversationBufferMemory# Initialize LLM 
llm = ChatOpenAI(model="gpt-4o-mini",temperature=0
)# Enhanced search tool
search = SerpAPIWrapper()
tools = [Tool(name="Search",func=search.run,description="Useful for searching latest AI research."),
]# Standard ReAct template (English)
react_template = """Answer the following questions as best you can. You have access to the following tools:{tools}Use the following format:Question: the input question you must answer
Thought: you should always think about what to do
Action: the action to take, should be one of [{tool_names}]
Action Input: the input to the action
Observation: the result of the action
... (this Thought/Action/Action Input/Observation can repeat N times)
Thought: I now know the final answer
Final Answer: the final answer to the original input questionConversation History:
{chat_history}Begin!Question: {input}
Thought:{agent_scratchpad}"""custom_prompt = PromptTemplate.from_template(react_template)# Create agent with memory
agent = create_react_agent(llm, tools, custom_prompt)
memory = ConversationBufferMemory(memory_key="chat_history",return_messages=True,input_key="input",output_key="output"
)agent_executor = AgentExecutor(agent=agent,tools=tools,memory=memory,verbose=True,handle_parsing_errors=True,max_iterations=3,return_intermediate_steps=False
)def clean_output(output):"""Extract final answer if exists"""if "Final Answer:" in output:return output.split("Final Answer:")[-1].strip()return output# First query
print("\n🔍 Query 1:")
result1 = agent_executor.invoke({"input": "What are the top 3 AI Agent research breakthroughs"})
answer1 = clean_output(result1["output"])
print("\n💡 Answer:", answer1)# Second query (with memory)
print("\n🔍 Query 2 (with memory):")
result2 = agent_executor.invoke({"input": "just list these 3 breakthroughs briefly",# Explicit memory injection"chat_history": [("human", result1["input"]),("ai", answer1)]
})
print("\n💡 Answer:", clean_output(result2["output"]))# Debug: Show full memory
print("\n🧠 Current Memory:")
print(memory.load_memory_variables({}))

测试 接口

from openai import OpenAIclient = OpenAI(api_key="sk-xxx28aD5",base_url = "https://xxxx"
)completion = client.chat.completions.create(model="gpt-4o-mini",store=True,messages=[{"role": "user", "content": "write a haiku about ai"}]
)print(completion.choices[0].message)

测试 图片模型

from openai import OpenAI
import os
from dotenv import load_dotenv
import requests# 加载环境变量
load_dotenv()# 初始化客户端（自动从环境变量读取OPENAI_API_KEY）
client = OpenAI(api_key="sk-xxx",base_url = "https://xxxxv1")try:response = client.images.generate(model="dall-e-3",  # 必须使用DALL·E模型prompt="电商生日礼物宣传海报",size="1024x1024",quality="standard",n=1,)# 安全获取URLif response.data and len(response.data) > 0:image_url = response.data[0].urlprint("图片生成成功，URL:", image_url)# 下载图片image_data = requests.get(image_url).contentwith open("flower_poster.png", "wb") as f:f.write(image_data)print("图片已保存为 flower_poster.png")else:print("错误：未收到有效的图片URL")except Exception as e:print(f"发生错误: {str(e)}")

参考书籍：大模型应用开发动手做AI Agent