📋Structured Outputs

Get guaranteed JSON responses that match your schema

Structured Outputs

When building applications, you need Claude to respond in a specific format you can parse. Structured outputs guarantee Claude's response matches your JSON schema.

The Problem Without Structured Outputs

JavaScript
// Without structured outputs — might get invalid JSON!
const response = await client.messages.create({
  model: "claude-sonnet-4-6",
  max_tokens: 1024,
  messages: [{
    role: "user",
    content: "Extract the name and email from: John Smith, john@example.com"
  }]
});
// Could return: "Name: John Smith, Email: john@example.com"
// Or: { "name": "John Smith" ... }
// Or anything else! 😱

The Solution: output_config

JavaScript
const response = await client.messages.create({
  model: "claude-sonnet-4-6",
  max_tokens: 1024,
  messages: [{
    role: "user",
    content: "Extract the name and email from: John Smith, john@example.com"
  }],
  output_config: {
    format: {
      type: "json_schema",
      schema: {
        type: "object",
        properties: {
          name: { type: "string" },
          email: { type: "string" }
        },
        required: ["name", "email"],
        additionalProperties: false
      }
    }
  }
});

// GUARANTEED to return:
// { "name": "John Smith", "email": "john@example.com" }

With SDK Helper (Recommended)

Python with Pydantic:

Python
from pydantic import BaseModel
from anthropic import Anthropic

class ContactInfo(BaseModel):
    name: str
    email: str
    company: str | None

client = Anthropic()
response = client.messages.parse(
    model="claude-sonnet-4-6",
    max_tokens=1024,
    output_format=ContactInfo,
    messages=[{
        "role": "user",
        "content": "Extract info: Sarah at Acme Corp, sarah@acme.com"
    }]
)

contact = response.parsed_output
print(contact.name)  # "Sarah"

TypeScript with Zod:

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

const ContactSchema = z.object({
  name: z.string(),
  email: z.string(),
  company: z.string().nullable()
});

const response = await client.messages.parse({
  model: "claude-sonnet-4-6",
  max_tokens: 1024,
  output_config: { format: zodOutputFormat(ContactSchema) },
  messages: [{ role: "user", content: "..." }]
});

Common Use Cases

Use Case	Schema Example
Data extraction	name, email, phone from text
Classification	category, confidence, tags
API responses	status, data, errors
Form parsing	structured fields from documents
Sentiment analysis	sentiment, score, keywords

Important Rules

additionalProperties must be false
No recursive schemas
Max 20 strict tools per request
First request has extra latency (grammar compilation)

Next up: Prompt engineering — the art of writing effective prompts.

Module 6

4/9

🔧 Tools & Web Search

Batch Processing 📦

4/9