免费在家构建智能体：开源工具与模型

#!/usr/bin/env python3
"""
使用Ollama + LangChain的本地研究智能体
前提条件：
  - ollama serve（在后台运行）
  - ollama pull mistral
  - pip install langchain langchain-ollama langchain-community duckduckgo-search
"""

import json
from langchain_core.tools import tool
from langchain_core.prompts import ChatPromptTemplate
from langchain_ollama import ChatOllama
from langchain.agents import AgentExecutor, create_react_agent
from duckduckgo_search import DDGS


@tool
def search_web(query: str) -> str:
    """使用DuckDuckGo在网络上搜索信息。"""
    try:
        with DDGS() as ddgs:
            results = list(ddgs.text(query, max_results=5))
        if not results:
            return "未找到结果。"
        return json.dumps(
            [{"title": r["title"], "snippet": r["body"], "url": r["href"]}
             for r in results],
            indent=2, ensure_ascii=False
        )
    except Exception as e:
        return f"搜索失败：{e}"


@tool
def read_file(filepath: str) -> str:
    """读取本地文件并返回其内容。"""
    try:
        with open(filepath, "r") as f:
            return f.read()
    except FileNotFoundError:
        return f"文件未找到：{filepath}"


@tool
def calculate(expression: str) -> str:
    """安全地评估简单的算术表达式。"""
    try:
        result = eval(expression, {"__builtins__": {}}, {})
        return str(result)
    except Exception as e:
        return f"计算错误：{e}"


llm = ChatOllama(
    model="mistral",
    base_url="http://localhost:11434",
    temperature=0.3,
)

prompt = ChatPromptTemplate.from_messages([
    ("system", """你是一个研究助手。使用你的工具回答问题。

可用工具：{tool_names}
工具描述：{tools}

严格遵循此格式：
Thought: 我需要找出什么？
Action: 工具名称
Action Input: 工具的输入
Observation: 工具结果
...（根据需要重复）
Thought: 现在我有足够的信息了。
Final Answer: 你的完整答案"""),
    ("user", "{input}"),
    ("placeholder", "{agent_scratchpad}"),
])

tools = [search_web, read_file, calculate]
agent = create_react_agent(llm, tools, prompt)
agent_executor = AgentExecutor(
    agent=agent,
    tools=tools,
    verbose=True,
    max_iterations=8,
    handle_parsing_errors=True,
)

if __name__ == "__main__":
    result = agent_executor.invoke({
        "input": "现在最流行的三个开源LLM是什么？给出名称和每个的关键特性。"
    })
    print(f"\n最终答案：\n{result['output']}")

import sqlite3

@tool
def query_database(sql: str) -> str:
    """对本地数据库执行只读SQL查询。"""
    try:
        conn = sqlite3.connect("data.db")
        cursor = conn.cursor()
        cursor.execute(sql)
        rows = cursor.fetchall()
        columns = [desc[0] for desc in cursor.description]
        conn.close()
        return json.dumps([dict(zip(columns, row)) for row in rows], indent=2)
    except Exception as e:
        return f"数据库错误：{e}"

from sentence_transformers import SentenceTransformer
import chromadb

model = SentenceTransformer("all-MiniLM-L6-v2")
client = chromadb.Client()
collection = client.create_collection("docs")

@tool
def search_docs(query: str) -> str:
    """通过语义相似度搜索本地文档。"""
    results = collection.query(
        query_embeddings=[model.encode(query).tolist()],
        n_results=3
    )
    return json.dumps(results["documents"][0], indent=2)

FROM python:3.12-slim
RUN pip install langchain langchain-ollama langchain-community duckduckgo-search
COPY research_agent.py .
ENV OLLAMA_HOST=host.docker.internal:11434
CMD ["python3", "research_agent.py"]

def run_pipeline(question: str) -> str:
    research_result = researcher.invoke({"input": question})
    final_output = writer.invoke({
        "input": f"基于此研究，写一个清晰的摘要：\n{research_result['output']}"
    })
    return final_output["output"]