Function Calling 与 MCP 实战

Function Calling 原理

Function Calling 让 LLM 能够调用外部函数/API。流程：

1. 你定义可用的函数（JSON Schema）
2. 用户提问
3. LLM 决定是否需要调用函数
4. 如果需要，LLM 返回函数名和参数
5. 你执行函数，返回结果
6. LLM 基于结果生成最终回答

Anthropic Tool Use vs OpenAI Function Calling

JSON Schema 定义工具

# Anthropic 格式
tool = {
    "name": "query_database",
    "description": "Query the user database",
    "input_schema": {
        "type": "object",
        "properties": {
            "sql": {
                "type": "string",
                "description": "SQL query to execute"
            },
            "limit": {
                "type": "integer",
                "description": "Max rows to return",
                "default": 10
            }
        },
        "required": ["sql"]
    }
}

# OpenAI 格式
function = {
    "name": "query_database",
    "description": "Query the user database",
    "parameters": {
        "type": "object",
        "properties": {
            "sql": {"type": "string"},
            "limit": {"type": "integer", "default": 10}
        },
        "required": ["sql"]
    }
}

实战：让 AI 操作数据库

import anthropic
import sqlite3

client = anthropic.Anthropic()
db = sqlite3.connect("users.db")

tools = [{
    "name": "query_db",
    "description": "Execute SQL query on user database",
    "input_schema": {
        "type": "object",
        "properties": {
            "sql": {"type": "string"}
        },
        "required": ["sql"]
    }
}]

def query_db(sql: str) -> str:
    cursor = db.execute(sql)
    rows = cursor.fetchall()
    return str(rows)

def chat(user_message: str):
    messages = [{"role": "user", "content": user_message}]

    response = client.messages.create(
        model="claude-sonnet-4-20250514",
        max_tokens=1024,
        tools=tools,
        messages=messages
    )

    if response.stop_reason == "tool_use":
        tool_use = next(b for b in response.content if b.type == "tool_use")
        result = query_db(tool_use.input["sql"])

        messages.append({"role": "assistant", "content": response.content})
        messages.append({
            "role": "user",
            "content": [{
                "type": "tool_result",
                "tool_use_id": tool_use.id,
                "content": result
            }]
        })

        final_response = client.messages.create(
            model="claude-sonnet-4-20250514",
            max_tokens=1024,
            messages=messages
        )
        return final_response.content[0].text

    return response.content[0].text

# 使用
print(chat("有多少个用户？"))
# AI 会生成 SQL: SELECT COUNT(*) FROM users

MCP 协议介绍

MCP（Model Context Protocol）是 Anthropic 推出的标准化工具协议，让 AI 应用能够连接各种数据源和工具。

MCP 架构

AI 应用（Claude Code / 你的 App）
  ↓ MCP Client
MCP Server（你实现的工具服务）
  ↓
外部系统（数据库 / API / 文件系统）

构建 MCP Server 步骤

1. 定义工具（Tools）
2. 实现工具逻辑
3. 启动 MCP Server
4. 在 AI 应用中配置 Server

示例：文件系统 MCP Server

from mcp.server import Server
from mcp.types import Tool, TextContent
import os

server = Server("filesystem")

@server.list_tools()
async def list_tools() -> list[Tool]:
    return [
        Tool(
            name="read_file",
            description="Read file contents",
            inputSchema={
                "type": "object",
                "properties": {
                    "path": {"type": "string"}
                },
                "required": ["path"]
            }
        ),
        Tool(
            name="list_dir",
            description="List directory contents",
            inputSchema={
                "type": "object",
                "properties": {
                    "path": {"type": "string"}
                },
                "required": ["path"]
            }
        )
    ]

@server.call_tool()
async def call_tool(name: str, arguments: dict) -> list[TextContent]:
    if name == "read_file":
        with open(arguments["path"]) as f:
            content = f.read()
        return [TextContent(type="text", text=content)]

    elif name == "list_dir":
        files = os.listdir(arguments["path"])
        return [TextContent(type="text", text="\n".join(files))]

# 启动 Server
server.run()

延伸阅读

• Anthropic Tool Use Guide
• MCP Documentation