System Prompt 与多轮对话

System Prompt

Python

import anthropic

client = anthropic.Anthropic(api_key="your-api-key", base_url="https://www.cheapertoken.work")

response = client.messages.create(
    model="claude-opus-4-7",
    max_tokens=1024,
    system="你是一个专业的Python编程助手，回答简洁准确。",
    messages=[
        {"role": "user", "content": "如何读取一个文件？"}
    ]
)
print(response.content[0].text)

JavaScript

import Anthropic from "@anthropic-ai/sdk";

const client = new Anthropic({ apiKey: "your-api-key", baseURL: "https://www.cheapertoken.work" });

const response = await client.messages.create({
  model: "claude-opus-4-7",
  max_tokens: 1024,
  system: "你是一个专业的Python编程助手，回答简洁准确。",
  messages: [{ role: "user", content: "如何读取一个文件？" }],
});
console.log(response.content[0].text);

curl

curl https://www.cheapertoken.work/v1/messages \
  -H "x-api-key: your-api-key" \
  -H "anthropic-version: 2023-06-01" \
  -H "content-type: application/json" \
  -d '{
    "model": "claude-opus-4-7",
    "max_tokens": 1024,
    "system": "你是一个专业的Python编程助手，回答简洁准确。",
    "messages": [{"role": "user", "content": "如何读取一个文件？"}]
  }'

多轮对话

Python

import anthropic

client = anthropic.Anthropic(api_key="your-api-key", base_url="https://www.cheapertoken.work")
messages = []

def chat(user_input):
    messages.append({"role": "user", "content": user_input})
    response = client.messages.create(
        model="claude-opus-4-7",
        max_tokens=1024,
        system="你是一个友好的助手。",
        messages=messages
    )
    reply = response.content[0].text
    messages.append({"role": "assistant", "content": reply})
    return reply

print(chat("我叫小明"))
print(chat("你还记得我叫什么吗？"))

JavaScript

import Anthropic from "@anthropic-ai/sdk";

const client = new Anthropic({ apiKey: "your-api-key", baseURL: "https://www.cheapertoken.work" });
const messages = [];

async function chat(userInput) {
  messages.push({ role: "user", content: userInput });
  const response = await client.messages.create({
    model: "claude-opus-4-7",
    max_tokens: 1024,
    system: "你是一个友好的助手。",
    messages,
  });
  const reply = response.content[0].text;
  messages.push({ role: "assistant", content: reply });
  return reply;
}

console.log(await chat("我叫小明"));
console.log(await chat("你还记得我叫什么吗？"));

curl

# 第一轮
curl https://www.cheapertoken.work/v1/messages \
  -H "x-api-key: your-api-key" \
  -H "anthropic-version: 2023-06-01" \
  -H "content-type: application/json" \
  -d '{
    "model": "claude-opus-4-7",
    "max_tokens": 1024,
    "system": "你是一个友好的助手。",
    "messages": [{"role": "user", "content": "我叫小明"}]
  }'

# 第二轮（携带历史）
curl https://www.cheapertoken.work/v1/messages \
  -H "x-api-key: your-api-key" \
  -H "anthropic-version: 2023-06-01" \
  -H "content-type: application/json" \
  -d '{
    "model": "claude-opus-4-7",
    "max_tokens": 1024,
    "system": "你是一个友好的助手。",
    "messages": [
      {"role": "user", "content": "我叫小明"},
      {"role": "assistant", "content": "你好，小明！很高兴认识你。"},
      {"role": "user", "content": "你还记得我叫什么吗？"}
    ]
  }'

多模态内容块

system 可以是字符串或内容块数组：

Python

import anthropic

client = anthropic.Anthropic(api_key="your-api-key", base_url="https://www.cheapertoken.work")

response = client.messages.create(
    model="claude-opus-4-7",
    max_tokens=1024,
    system=[
        {"type": "text", "text": "你是代码助手。"},
        {"type": "text", "text": "始终提供可运行的代码示例。"}
    ],
    messages=[{"role": "user", "content": "写一个快速排序"}]
)
print(response.content[0].text)

JavaScript

import Anthropic from "@anthropic-ai/sdk";

const client = new Anthropic({ apiKey: "your-api-key", baseURL: "https://www.cheapertoken.work" });

const response = await client.messages.create({
  model: "claude-opus-4-7",
  max_tokens: 1024,
  system: [
    { type: "text", text: "你是代码助手。" },
    { type: "text", text: "始终提供可运行的代码示例。" },
  ],
  messages: [{ role: "user", content: "写一个快速排序" }],
});
console.log(response.content[0].text);

curl

curl https://www.cheapertoken.work/v1/messages \
  -H "x-api-key: your-api-key" \
  -H "anthropic-version: 2023-06-01" \
  -H "content-type: application/json" \
  -d '{
    "model": "claude-opus-4-7",
    "max_tokens": 1024,
    "system": [
      {"type": "text", "text": "你是代码助手。"},
      {"type": "text", "text": "始终提供可运行的代码示例。"}
    ],
    "messages": [{"role": "user", "content": "写一个快速排序"}]
  }'