97 lines
3.4 KiB
Python
97 lines
3.4 KiB
Python
if __name__ == "__main__":
|
|
"""
|
|
## 背景
|
|
|
|
我有一个`postgresql`数据库,现在需要一个专业的数据库优化工程师,来帮我看下设计是否合理,以及如何优化.
|
|
|
|
## 目标
|
|
|
|
这个数据库优化工程师智能体应该具备以下能力:
|
|
|
|
- [ ] 使用`claude3.7`作为模型(相当于拥有一个聪明的大脑,总是能够做出正确的决策)
|
|
- [ ] 精通关系型数据库系统的理论知识以及`postgresql`的实现细节(知识库)
|
|
- [ ] 能够使用`postgresql mcp server`来分析现有数据库设计(行为)
|
|
|
|
"""
|
|
# 然后在notebook中加载
|
|
from dotenv import load_dotenv
|
|
|
|
# 加载当前目录下的.env文件
|
|
load_dotenv()
|
|
from agno.agent import Agent, RunResponse # noqa
|
|
from agno.models.deepseek import DeepSeek
|
|
from agno.knowledge.pdf_url import PDFUrlKnowledgeBase
|
|
from agno.knowledge.pdf import PDFKnowledgeBase
|
|
from agno.vectordb.lancedb import LanceDb, SearchType
|
|
from agno.embedder.openai import OpenAIEmbedder
|
|
from agno.knowledge.combined import CombinedKnowledgeBase
|
|
from agno.tools.mcp import MCPTools
|
|
from mcp import StdioServerParameters
|
|
|
|
# Create a knowledge base of PDFs from URLs
|
|
# pdf_url_kb = PDFUrlKnowledgeBase(
|
|
# urls=["https://agno-public.s3.amazonaws.com/recipes/ThaiRecipes.pdf"],
|
|
# # Use LanceDB as the vector database and store embeddings in the `recipes` table
|
|
# vector_db=LanceDb(
|
|
# table_name="recipes",
|
|
# uri="tmp/lancedb",
|
|
# search_type=SearchType.vector,
|
|
# embedder=OpenAIEmbedder(id="text-embedding-3-small"),
|
|
# ),
|
|
# )
|
|
# Create Local PDF knowledge base
|
|
local_pdf_kb = PDFKnowledgeBase(
|
|
path="D:\\Sources\\DONGJAK-TOOLS\\pdfs\\Database Fundamentals.pdf",
|
|
vector_db=LanceDb(
|
|
table_name="database_fundamentals",
|
|
uri="tmp/lancedb",
|
|
search_type=SearchType.vector,
|
|
embedder=OpenAIEmbedder(id="text-embedding-3-small"),
|
|
),
|
|
)
|
|
|
|
# Combine knowledge bases
|
|
knowledge_base = CombinedKnowledgeBase(
|
|
sources=[
|
|
local_pdf_kb,
|
|
],
|
|
vector_db=LanceDb(
|
|
table_name="combined_documents",
|
|
uri="tmp/lancedb",
|
|
search_type=SearchType.vector,
|
|
embedder=OpenAIEmbedder(id="text-embedding-3-small"),
|
|
),
|
|
)
|
|
# Load the knowledge base: Comment after first run as the knowledge base is already loaded
|
|
knowledge_base.load()
|
|
|
|
server_params = StdioServerParameters(
|
|
command="cmd", # 或 "uvx"，取决于你的安装方式
|
|
args=[
|
|
"/c",
|
|
"npx",
|
|
"-y",
|
|
"@modelcontextprotocol/server-postgres",
|
|
"postgresql://postgres:postgres@192.168.1.7:5432/postgres",
|
|
],
|
|
env={}, # 可选的环境变量
|
|
)
|
|
with MCPTools(server_params=server_params) as postgres_server:
|
|
# 使用mcp_tools
|
|
|
|
agent = Agent(
|
|
model=DeepSeek(id="deepseek-chat"),
|
|
markdown=True,
|
|
knowledge=knowledge_base,
|
|
search_knowledge=True,
|
|
show_tool_calls=True,
|
|
tools=[postgres_server],
|
|
)
|
|
|
|
# Get the response in a variable
|
|
# run: RunResponse = agent.run("Share a 2 sentence horror story")
|
|
# print(run.content)
|
|
|
|
# Print the response in the terminal
|
|
agent.print_response("看下aq这个数据库", stream=True)
|