Claude 预填充（prefill）400 报错：4.6 起模型全拒收，按用途替换

# 以 assistant 消息结尾 = 预填充
response = client.messages.create(
    model="claude-haiku-4-5-20251001",  # Haiku 4.5 仍接受
    max_tokens=1024,
    messages=[
        {"role": "user", "content": "List the top 3 programming languages"},
        {"role": "assistant", "content": "{"},  # 预填充
    ],
)

{
  "type": "error",
  "error": {
    "type": "invalid_request_error",
    "message": "This model does not support assistant message prefill. The conversation must end with a user message."
  }
}

/claude-api migrate this project to claude-opus-5-5

import anthropic

client = anthropic.Anthropic()

response = client.messages.create(
    model="claude-sonnet-5-5",
    max_tokens=1024,
    messages=[
        {"role": "user", "content": "List the top 3 programming languages by popularity"},
        {"role": "assistant", "content": "{"},  # 400
    ],
)

import json
import anthropic

client = anthropic.Anthropic()

response = client.messages.create(
    model="claude-sonnet-5-5",
    max_tokens=1024,
    messages=[
        {"role": "user", "content": "List the top 3 programming languages by popularity"}
    ],
    output_config={
        "format": {
            "type": "json_schema",
            "schema": {
                "type": "object",
                "properties": {
                    "languages": {
                        "type": "array",
                        "items": {
                            "type": "object",
                            "properties": {
                                "name": {"type": "string"},
                                "rank": {"type": "integer"},
                            },
                            "required": ["name", "rank"],
                            "additionalProperties": False,
                        },
                    }
                },
                "required": ["languages"],
                "additionalProperties": False,
            },
        }
    },
)

# 按类型取文本块：新模型的回复里 thinking 块可能排在前面
text = next(block.text for block in response.content if block.type == "text")
data = json.loads(text)
print(data["languages"][0]["name"])

from pydantic import BaseModel
import anthropic

client = anthropic.Anthropic()

class Language(BaseModel):
    name: str
    rank: int

class LanguageList(BaseModel):
    languages: list[Language]

response = client.messages.parse(
    model="claude-sonnet-5-5",
    max_tokens=1024,
    messages=[
        {"role": "user", "content": "List the top 3 programming languages by popularity"}
    ],
    output_format=LanguageList,  # SDK 辅助参数，发出去的是 output_config.format
)

for lang in response.parsed_output.languages:
    print(lang.rank, lang.name)

import Anthropic from "@anthropic-ai/sdk";
import { z } from "zod";
import { zodOutputFormat } from "@anthropic-ai/sdk/helpers/zod";

const client = new Anthropic();

const LanguageList = z.object({
  languages: z.array(z.object({ name: z.string(), rank: z.number() })),
});

const response = await client.messages.parse({
  model: "claude-sonnet-5-5",
  max_tokens: 1024,
  messages: [
    { role: "user", content: "List the top 3 programming languages by popularity" },
  ],
  output_config: { format: zodOutputFormat(LanguageList) },
});

console.log(response.parsed_output);

curl https://api.anthropic.com/v1/messages \
  -H "content-type: application/json" \
  -H "x-api-key: $ANTHROPIC_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -d '{
    "model": "claude-sonnet-5-5",
    "max_tokens": 1024,
    "messages": [
      {"role": "user", "content": "List the top 3 programming languages by popularity"}
    ],
    "output_config": {
      "format": {
        "type": "json_schema",
        "schema": {
          "type": "object",
          "properties": {
            "languages": {
              "type": "array",
              "items": {
                "type": "object",
                "properties": {
                  "name": {"type": "string"},
                  "rank": {"type": "integer"}
                },
                "required": ["name", "rank"],
                "additionalProperties": false
              }
            }
          },
          "required": ["languages"],
          "additionalProperties": false
        }
      }
    }
  }'

response = client.messages.create(
    model="claude-sonnet-5-5",
    max_tokens=1024,
    messages=[
        {"role": "user", "content": "Classify this ticket: 'I was charged twice this month.'"}
    ],
    tools=[
        {
            "name": "record_category",
            "description": "Record the category of a support ticket",
            "strict": True,
            "input_schema": {
                "type": "object",
                "properties": {
                    "category": {
                        "type": "string",
                        "enum": ["billing", "bug", "feature_request", "other"],
                    }
                },
                "required": ["category"],
                "additionalProperties": False,
            },
        }
    ],
    # tool_choice 保持默认 auto
)

for block in response.content:
    if block.type == "tool_use":
        print(block.input["category"])

response = client.messages.create(
    model="claude-sonnet-5-5",
    max_tokens=1024,
    messages=[
        {"role": "user", "content": "Analyze this code for bugs"},
        {"role": "assistant", "content": "## Bug Analysis\n\n"},  # 400
    ],
)

response = client.messages.create(
    model="claude-sonnet-5-5",
    max_tokens=1024,
    system=(
        "You are a code review assistant. Answer directly, without a preamble. "
        "Start with a '## Bug Analysis' heading, list each bug with its severity, "
        "then give a fix for each one."
    ),
    messages=[
        {"role": "user", "content": "Analyze this code for bugs"}
    ],
)

question = "Write a migration plan for our billing service"

first = client.messages.create(
    model="claude-sonnet-5-5",
    max_tokens=1024,
    messages=[{"role": "user", "content": question}],
)
partial = "".join(b.text for b in first.content if b.type == "text")

if first.stop_reason == "max_tokens":
    rest = client.messages.create(
        model="claude-sonnet-5-5",
        max_tokens=4096,
        messages=[
            {
                "role": "user",
                "content": [
                    {"type": "text", "text": question},
                    {
                        "type": "text",
                        "text": (
                            "Your previous response was interrupted and ended with:\n\n"
                            f"{partial}\n\n"
                            "Continue from where you left off."
                        ),
                    },
                ],
            }
        ],
    )
    continuation = "".join(b.text for b in rest.content if b.type == "text")
    full_text = partial + continuation  # 拼接前检查接缝处有没有重复

def ensure_user_tail(messages):
    """发请求前确认最后一条是 user 消息。"""
    msgs = list(messages)

    # 去掉末尾内容为空的 assistant 轮次
    while msgs and msgs[-1]["role"] == "assistant" and not msgs[-1].get("content"):
        msgs.pop()

    if not msgs or msgs[-1]["role"] != "user":
        # 末尾仍是有内容的 assistant 消息：这是续写或预填充，
        # 要按用途改写，而不是静默删掉
        raise ValueError("messages must end with a user message")
    return msgs