Mengisi awal respons Claude untuk kontrol output yang lebih baik

While these tips apply broadly to all Claude models, you can find prompting tips specific to extended thinking models here.

Pengisian awal hanya tersedia untuk mode pemikiran non-extended. Saat ini tidak didukung dengan pemikiran extended.

Saat menggunakan Claude, Anda memiliki kemampuan unik untuk mengarahkan responsnya dengan mengisi awal pesan Assistant. Teknik yang ampuh ini memungkinkan Anda mengarahkan tindakan Claude, melewati pendahuluan, memaksakan format tertentu seperti JSON atau XML, dan bahkan membantu Claude mempertahankan konsistensi karakter dalam skenario bermain peran.

Dalam beberapa kasus di mana Claude tidak bekerja sesuai harapan, beberapa kalimat yang diisi di awal dapat sangat meningkatkan kinerja Claude. Sedikit pengisian awal bisa memberikan dampak besar!

Cara mengisi awal respons Claude

Untuk mengisi awal, sertakan teks awal yang diinginkan dalam pesan Assistant (respons Claude akan melanjutkan dari tempat pesan Assistant berakhir):

import anthropic

client = anthropic.Anthropic()
response = client.messages.create(
    model="claude-3-7-sonnet-20250219",
    max_tokens=1024,
    messages=[
        {"role": "user", "content": "What is your favorite color?"},
        {"role": "assistant", "content": "As an AI assistant, I don't have a favorite color, But if I had to pick, it would be green because"}  # Prefill here
    ]
)

Contoh

Contoh 1: Mengontrol format output dan melewati pendahuluan

Tip pengguna ahli: Mengisi awal { memaksa Claude untuk melewati pendahuluan dan langsung menghasilkan objek JSON. Ini lebih bersih, lebih ringkas, dan lebih mudah diurai oleh program tanpa pemrosesan tambahan.

Role

Content

User

Extract the name, size, price, and color from this product description as a JSON object:

<description>
The SmartHome Mini is a compact smart home assistant available in black or white for only $49.99. At just 5 inches wide, it lets you control lights, thermostats, and other connected devices via voice or app—no matter where you place it in your home. This affordable little hub brings convenient hands-free control to your smart devices.
</description>

Assistant

Here’s the extracted information in JSON format:

```json
{
“name”: “SmartHome Mini”,
“size”: “5 inches wide”,
“price”: “$49.99”,
“colors”: [“black”, “white”]
}
```

I’ve extracted the following details from the product description:
- Name: SmartHome Mini
- Size: 5 inches wide
- Price: $49.99
- Colors: Available in black and white

The JSON object provides a structured format that’s easy for programs to parse and use. Let me know if you need any modifications or have any other questions!

Role	Content
User	Extract the name, size, price, and color from this product description as a JSON object: <description> The SmartHome Mini is a compact smart home assistant available in black or white for only $49.99. At just 5 inches wide, it lets you control lights, thermostats, and other connected devices via voice or app—no matter where you place it in your home. This affordable little hub brings convenient hands-free control to your smart devices. </description>
Assistant (prefill)	{
Assistant (Claude’s response)	“name”: “SmartHome Mini”, “size”: “5 inches wide”, “price”: “$49.99”, “colors”: [ “black”, “white” ] }

Contoh 2: Mempertahankan karakter dalam skenario bermain peran

Tip bermain peran: Mengisi awal [ROLE_NAME] dapat mengingatkan Claude untuk tetap dalam karakter, bahkan untuk percakapan yang lebih panjang dan kompleks. Ini sangat ampuh ketika dikombinasikan dengan prompt peran dalam parameter system.

Role	Content
User (50 conversation turns in with “Sherlock Holmes” Claude)	What do you deduce about the owner of this shoe? <image>muddy_shoe.jpg</image>
Assistant	As an AI assistant, I would be happy to analyze the image of a muddy shoe as if you had provided it. Here’s my deduction…

Role	Content
User (50 conversation turns in with “Sherlock Holmes” Claude)	What do you deduce about the owner of this shoe? <image>muddy_shoe.jpg</image>
Assistant (prefill)	[Sherlock Holmes]
Assistant (Claude’s response)	Ah, what have we here? A most curious specimen, Watson! Let us examine this shoe with a discerning eye. The mud, you see, is not your common London clay, but rather the distinctive reddish hue of the iron-rich soil found only in the environs of Hampstead Heath. And yet, the wear on the sole suggests a gait more suited to the cobblestones of Whitechapel than the gentle slopes of the Heath. Now, observe the laces! They are tied with a particular knot favored by those in the medical profession—doubtless a habit acquired during years of tying surgical gowns. But the scuff marks on the toe box tell a different tale: they are consistent with the kicking motion one might employ in the martial arts, specifically Baritsu. We may conclude, then, that our absent shoe owner is a doctor, possibly a surgeon, who resides in Whitechapel but frequents Hampstead Heath—perhaps for a secret Baritsu practice, away from prying eyes. A most intriguing case, Watson! We must make haste; the game is afoot!

Perpustakaan prompt

Dapatkan inspirasi dari pilihan prompt yang dikurasi untuk berbagai tugas dan kasus penggunaan.

Tutorial prompting GitHub

Tutorial yang penuh contoh yang mencakup konsep-konsep prompt engineering yang ada dalam dokumentasi kami.

Tutorial prompting Google Sheets

Versi yang lebih ringan dari tutorial prompt engineering kami melalui spreadsheet interaktif.

Mulai

Agen dan alat

Pelajari tentang Claude

Bangun dengan Claude

Administrasi

Uji dan evaluasi

Sumber Daya

Pusat Legal

Mengisi awal respons Claude untuk kontrol output yang lebih baik

Cara mengisi awal respons Claude

Contoh

Contoh 1: Mengontrol format output dan melewati pendahuluan

Contoh 2: Mempertahankan karakter dalam skenario bermain peran

Perpustakaan prompt

Tutorial prompting GitHub

Tutorial prompting Google Sheets

Mulai

Agen dan alat

Pelajari tentang Claude

Bangun dengan Claude

Administrasi

Uji dan evaluasi

Sumber Daya

Pusat Legal

​Cara mengisi awal respons Claude

​Contoh

​Contoh 1: Mengontrol format output dan melewati pendahuluan

​Contoh 2: Mempertahankan karakter dalam skenario bermain peran

Perpustakaan prompt

Tutorial prompting GitHub

Tutorial prompting Google Sheets

Cara mengisi awal respons Claude

Contoh

Contoh 1: Mengontrol format output dan melewati pendahuluan

Contoh 2: Mempertahankan karakter dalam skenario bermain peran