CopilotKit 仓库中的coagents-starter示例项目
langchain-openai = ">=0.3.9,<0.4.0"# 确保只有这一个约束。# 用记事本编辑（或使用其他编辑器如VS Code）打开一个新的终端窗口，进入 ui 目录并启动前端服务。访问端口：http://localhost:3000。修改 pyproject.toml 强制统一版本。：使用 Python 执行下载好的脚本。Python3.7以上。下载脚本到本地文件。
环境：
pnpm：可以使用 npm 进行安装，命令为 npm install -g pnpm。
Python3.7以上
- Poetry 环境安装：下载脚本：使用 Invoke-WebRequest 下载脚本到本地文件。
Invoke-WebRequest -Uri https://install.python-poetry.org -OutFile install-poetry.py
运行脚本：使用 Python 执行下载好的脚本。
python install-poetry.py
手动更新 PATH 环境变量
你可以按照以下步骤把 C:\Users\24558\AppData\Roaming\Python\Scripts 添加到系统的 PATH 环境变量中：
- 按下 Win + R 组合键，打开 “运行” 对话框，输入 sysdm.cpl 并回车，打开 “系统属性” 窗口。
- 在 “系统属性” 窗口中，切换到 “高级” 选项卡，点击 “环境变量” 按钮。
- 在 “环境变量” 窗口中，找到 “系统变量” 列表中的 Path 变量，选中它并点击 “编辑” 按钮。
- 在 “编辑环境变量” 窗口中，点击 “新建” 按钮，将 C:\Users\24558\AppData\Roaming\Python\Scripts 添加到列表中。
- 依次点击 “确定” 保存更改。
- 克隆仓库
cd CopilotKit/examples/coagents-starter
- 进入 ui 目录并安装前端依赖
cd ui
pnpm install
- 安装后端依赖（Python 代理）
cd ../agent-py
poetry install
- 运行后端服务（Python 代理）：poetry run uvicorn sample_agent.demo:app --reload
- 运行前端服务
打开一个新的终端窗口，进入 ui 目录并启动前端服务
cd ui
pnpm run dev
访问端口：http://localhost:3000
关键技术难点：
1.修改调整前端demo.py代码：
"""
This serves the "sample_agent" agent.
"""
from langgraph.checkpoint.base import CheckpointMetadata
EXCLUDED_METADATA_KEYS = getattr(CheckpointMetadata, "EXCLUDED_KEYS", set())
import os
from dotenv import load_dotenv
load_dotenv()
from fastapi import FastAPI
import uvicorn
from copilotkit.integrations.fastapi import add_fastapi_endpoint
from copilotkit import CopilotKitRemoteEndpoint, LangGraphAgent
from sample_agent.agent import graph
from langgraph.checkpoint.memory import MemorySaver
app = FastAPI()
# 创建检查点实例
checkpointer = MemorySaver()
# 正确配置代理（通过graph配置检查点）
graph.checkpointer = checkpointer # 直接在graph上设置检查点
sdk = CopilotKitRemoteEndpoint(
agents=[
LangGraphAgent(
name="sample_agent",
description="An example agent",
graph=graph # 只传递graph参数
)
],
)
add_fastapi_endpoint(app, sdk, "/copilotkit")
def main():
port = int(os.getenv("PORT", "8000"))
uvicorn.run(
"sample_agent.demo:app",
host="0.0.0.0",
port=port,
reload=True,
)
if __name__ == "__main__":
main()
2.修改调整后端agent.py代码（设置为调用DeepSeek API服务）
"""
This is the main entry point for the agent.
It defines the workflow graph, state, tools, nodes and edges.
"""
from dotenv import load_dotenv
import os
from typing_extensions import Literal
from langchain_openai import ChatOpenAI
from langchain_core.messages import SystemMessage, AIMessage
from langchain_core.runnables import RunnableConfig
from langchain.tools import tool
from langgraph.graph import StateGraph, END
from langgraph.types import Command
from langgraph.prebuilt import ToolNode
from copilotkit import CopilotKitState
load_dotenv()
class AgentState(CopilotKitState):
"""
Here we define the state of the agent
In this instance, we're inheriting from CopilotKitState, which will bring in
the CopilotKitState fields. We're also adding a custom field, `language`,
which will be used to set the language of the agent.
"""
proverbs: list[str] = []
# your_custom_agent_state: str = ""
@tool
def get_weather(location: str):
"""
Get the weather for a given location.
"""
return f"The weather for {location} is 70 degrees."
# @tool
# def your_tool_here(your_arg: str):
# """Your tool description here."""
# print(f"Your tool logic here")
# return "Your tool response here."
tools = [
get_weather
# your_tool_here
]
async def chat_node(state: AgentState, config: RunnableConfig):
"""
Standard chat node based on the ReAct design pattern. It handles:
- The model to use (and binds in CopilotKit actions and the tools defined above)
- The system prompt
- Getting a response from the model
- Handling tool calls
For more about the ReAct design pattern, see:
https://www.perplexity.ai/search/react-agents-NcXLQhreS0WDzpVaS4m9Cg
"""
# 1. Define the model
model = ChatOpenAI(
openai_api_key=os.getenv("DEEPSEEK_API_KEY"),
openai_api_base="https://api.deepseek.com/v1",
model_name="deepseek-chat",
temperature=0.7,
max_tokens=2048,
)
# 2. Bind the tools to the model
model_with_tools = model.bind_tools(
[
*state["copilotkit"]["actions"],
get_weather,
# your_tool_here
],
# 2.1 Disable parallel tool calls to avoid race conditions,
# enable this for faster performance if you want to manage
# the complexity of running tool calls in parallel.
parallel_tool_calls=False,
)
# 3. Define the system message by which the chat model will be run
system_message = SystemMessage(
content=f"You are a helpful assistant. Talk in {state.get('language', 'english')}."
)
# 4. Run the model to generate a response
response = await model_with_tools.ainvoke([
system_message,
*state["messages"],
], config)
# 5. Check for tool calls in the response and handle them. We ignore
# CopilotKit actions, as they are handled by CopilotKit.
if isinstance(response, AIMessage) and response.tool_calls:
actions = state["copilotkit"]["actions"]
# 5.1 Check for any non-copilotkit actions in the response and
# if there are none, go to the tool node.
if not any(
action.get("name") == response.tool_calls[0].get("name")
for action in actions
):
return Command(goto="tool_node", update={"messages": response})
# 6. We've handled all tool calls, so we can end the graph.
return Command(
goto=END,
update={
"messages": response
}
)
# Define the workflow graph
workflow = StateGraph(AgentState)
workflow.add_node("chat_node", chat_node)
workflow.add_node("tool_node", ToolNode(tools=tools))
workflow.add_edge("tool_node", "chat_node")
workflow.set_entry_point("chat_node")
# Compile the workflow graph
graph = workflow.compile()
3.langchain-openai 依赖冲突错误：
修改 pyproject.toml 强制统一版本
# 用记事本编辑（或使用其他编辑器如VS Code）
notepad pyproject.toml
[tool.poetry.dependencies]
python = "^3.8"
langchain-openai = ">=0.3.9,<0.4.0" # 确保只有这一个约束
欢迎加入 MCP 技术社区！与志同道合者携手前行，一同解锁 MCP 技术的无限可能！
更多推荐
48
0- 0
已为社区贡献1条内容
所有评论(0)