@font-face {
  font-family: 'Styrene Display';
  src: url('https://www-cdn.anthropic.com/e8f9c8ca51b03efb6315db351446fc972ab15abe/StyreneA-Medium-Web.woff2') format('woff2');
  font-weight: 500;
  font-style: normal;
  font-stretch: normal;
}

@font-face {
  font-family: 'Styrene Display';
  src: url('https://www-cdn.anthropic.com/e8f9c8ca51b03efb6315db351446fc972ab15abe/StyreneA-Medium-Web.woff2') format('woff2');
  font-weight: 600;
  font-style: normal;
  font-stretch: normal;
}

@font-face {
  font-family: 'Styrene';
  src: url('https://www-cdn.anthropic.com/6f87b6d99aefde021ac24f21295bf9e70f71472f/StyreneBLC-Regular.woff2') format('woff2');
  font-weight: 400;
  font-style: normal;
  font-stretch: normal;
}


@font-face {
  font-family: 'Styrene';
  src: url('https://www-cdn.anthropic.com/6f87b6d99aefde021ac24f21295bf9e70f71472f/StyreneBLC-Regular.woff2') format('woff2');
  font-weight: 500;
  font-style: normal;
  font-stretch: normal;
}
 
 @font-face {
  font-family: 'Styrene';
  src: url('https://www-cdn.anthropic.com/3611e9e4aaaf466dbd47e2686f561e7de694cb6c/StyreneBLC-Medium.woff2') format('woff2');
  font-weight: 600;
  font-style: normal;
  font-stretch: normal;
}
 

@font-face {
  font-family: 'Tiempos';
  src: url('https://www-cdn.anthropic.com/c3e09cefbfeb4e5eaca56b7bc8b9a1aa1aeda025/TiemposText-Regular.woff2') format('woff2');
  font-weight: 400;
  font-style: normal;
  font-stretch: normal;
}

@font-face {
  font-family: 'Tiempos';
  src: url('https://www-cdn.anthropic.com/b198ca4e31a323b2abb84b3eeeb1eed1f471afa0/TiemposText-Medium.woff2') format('woff2');
  font-weight: 500;
  font-style: normal;
  font-stretch: normal;
}

@font-face {
  font-family: 'Tiempos';
  src: url('https://www-cdn.anthropic.com/b198ca4e31a323b2abb84b3eeeb1eed1f471afa0/TiemposText-Medium.woff2') format('woff2');
  font-weight: 600;
  font-style: normal;
  font-stretch: normal;
}

body, input, #category-select, .dropdown-item, #table-of-contents {
  font-family: 'Styrene', sans-serif;
}

.eyebrow {
  font-family: 'Styrene Display', sans-serif;
  text-transform: uppercase;
  letter-spacing: .02rem;
}

#content-container {
  font-family: 'Tiempos', serif;
}

#content-container h1, #content-container h2, #content-container h3, #content-container h4, #content-container h5, #content-container h6 {
  font-family: 'Styrene Display', sans-serif;
}

#content-container p {
  font-size: 1rem;
  line-height: 1.65rem;
}

.font-extrabold {
  font-weight: 600 !important;
}

.wide-table {
  width: 100%;
  overflow-x: auto;
}

.wide-table table {
  width: 175%;
  margin-bottom: 0;
}

/* Prompt Library */
#prompt-library-container {
  margin: 4rem auto;
  max-width: 48rem;
  padding-left: 1.25rem;
  padding-right: 1.25rem;
}

.prompt-library-title {
  font-size: 24px;
  text-align: center;
  font-weight: 700;
  color: #1f2937;
}

.dark .prompt-library-title {
  color: #e5e7eb;
}

.prompt-library-description {
  margin-top: 1rem;
  text-align: center;
}

.main-content {
  margin-bottom: 10rem;
  max-width: 64rem;
  margin-left: auto;
  margin-right: auto;
  padding-left: 1.25rem;
  padding-right: 1.25rem;
}

.prompt-controllers {
  display: flex;
  gap: 0.5rem;
}

.prompt-search-container {
  position: relative;
  flex: 1 1 0%;
}

.prompt-search-icon-container {
  display: flex;
  position: absolute;
  top: 0;
  bottom: 0;
  left: 0;
  align-items: center;
  padding-left: 0.75rem;
}

.prompt-search-icon {
  margin-left: 0.25rem;
  margin-right: 0.75rem;
  flex: none;
  width: 1rem;
  height: 1rem;
  background-color: #6b7280;
  mask-image: url(https://mintlify.b-cdn.net/v6.5.1/solid/magnifying-glass.svg);
  mask-repeat: no-repeat;
  mask-position: center center;
}

input.prompt-search-bar {
  display: block;
  height: 2.5rem;
  padding-left: 2.5rem;
  border-radius: 0.75rem;
  border-width: 1px;
  background-color: #ffffff;
  width: 100%;
  color: #111827;
  box-shadow: 0 1px 2px 0 rgba(0, 0, 0, 0.05);
}

.dark input.prompt-search-bar {
  color: #ffffff;
  background-color: rgb(var(--background-dark));
  border-color: #d1d5db1a;
}

input.prompt-search-bar:focus {
  outline-color: rgb(var(--primary));
}

.dark input.prompt-search-bar:focus {
  outline-color: rgb(var(--primary-light));
}

.dark .prompt-search-icon {
  background-color: #ffffff80;
}

#category-select {
  padding-left: 1rem;
  padding-right: 2.5rem;
  height: 2.5rem;
  display: flex;
  align-items: center;
  border-radius: 0.75rem;
  border-width: 1px;
  color: #111827;
  background-color: #ffffff;
  cursor: pointer;
  box-shadow: 0 1px 2px 0 rgba(0, 0, 0, 0.05);
  white-space: nowrap;
}

.dark #category-select {
  background-color: rgb(var(--background-dark));
  border-color: #d1d5db1a;
  color: #ffffff;
}

#category-select:hover {
  background-color: #f9fafb;
}

.dark #category-select:hover {
  background-color: #ffffff0d;
}

#category-select:focus {
  outline-color: rgb(var(--primary));
}

.dark #category-select:focus {
  outline-color: rgb(var(--primary-light));
}

#categories-dropdown {
  top: calc(100% + 4px);
  padding: 0.5rem 0.5rem;
  display: none;
  position: absolute;
  z-index: 10;
  border-radius: 0.75rem;
  border-width: 1px;
  width: 100%;
  color: #111827;
  background-color: #ffffff;
  box-shadow: 0 1px 2px 0 rgba(0, 0, 0, 0.05);
}

.dark #categories-dropdown {
  background-color: rgb(var(--background-dark));
  border-color: #d1d5db1a;
  color: #ffffff;
}

#categories-dropdown-clickout {
  position: fixed;
  top: 0;
  right: 0;
  bottom: 0;
  left: 0;
  z-index: 0;
}

.dropdown-icon-container {
  display: flex;
  position: absolute;
  top: 0;
  bottom: 0;
  right: 0;
  align-items: center;
  padding-right: 0.25rem;
}

.dropdown-icon {
  margin-left: 0.25rem;
  margin-right: 0.75rem;
  flex: none;
  width: 0.75rem;
  height: 0.75rem;
  background-color: #6b7280;
  mask-image: url(https://mintlify.b-cdn.net/v6.5.1/solid/caret-down.svg);
  mask-repeat: no-repeat;
  mask-position: center center;
}

.dark .dropdown-icon {
  background-color: #ffffff80;
}

#prompts-container {
  grid-template-columns: repeat(1, minmax(0, 1fr));
  gap: 2rem;
}

.dropdown-item {
  padding: 0.25rem 0.5rem;
  border-radius: 0.375rem;
  display: flex;
  align-items: center;
  cursor: pointer;
}

.dropdown-item:hover {
  background-color: #f9fafb;
}

.dark .dropdown-item:hover {
  background-color: #ffffff0d;
}

.check-icon {
  mask-image: url(https://mintlify.b-cdn.net/v6.5.1/solid/check.svg);
  height: 0.875rem;
  width: 1rem;
  background-color: rgb(var(--primary-light));
  mask-repeat: no-repeat;
  mask-position: center center;
}

.prompt-card {
  margin: -0.75rem;
  padding: 0.75rem;
  display: flex;
  border-radius: 1rem;
}

.prompt-card:hover {
  background-color: #03071208;
}

.dark .prompt-card:hover {
  background-color: #ffffff08;
}

.prompt-icon-container {
  display: flex;
  flex: none;
  align-items: center;
  justify-content: center;
  margin-right: 1.5rem;
  border-radius: 0.75rem;
  height: 4rem;
  width: 4rem;
  background-color: #cb785c1a;
}

.prompt-icon {
  height: 1.5rem;
  width: 1.5rem;
  background-color: rgb(var(--primary-light));
  mask-repeat: no-repeat;
  mask-position: center center;
}

.prompt-title {
  color: rgb(31 41 55);
  font-weight: 600;
}

.dark .prompt-title {
  color: rgb(229 231 235);
}

.prompt-description {
  margin-top: 0.25rem;
}

#prompts-container {
  display: grid;
  margin-top: 2.5rem;
}

@media (min-width: 640px) {
  #category-select {
    width: 16rem;
  }
}

@media (min-width: 1024px) {
  #prompts-container {
    grid-template-columns: repeat(2, minmax(0, 1fr));
  }
}

/* Utility classes */
.relative {
  position: relative;
}

.flex-1 {
  flex: 1 1 0%;
}


.prompts-container {
  grid-template-columns: repeat(1, minmax(0, 1fr));
  gap: 2rem;
}

@media (min-width: 1024px) {
  .prompts-container {
    grid-template-columns: repeat(2, minmax(0, 1fr));
  }
}


Developer Console

Developer Discord

Support

Console Développeur

Discord pour les développeurs

Assistance

Entwicklerkonsole

Entwickler-Discord

Unterstützung

Console per sviluppatori

Discord per Sviluppatori

Supporto

開発者コンソール

開発者向けDiscord

サポート

Consola de Desarrollador

Discord para Desarrolladores

Soporte

개발자 콘솔

개발자 Discord

지원

开发者控制台

开发者 Discord

支持

開發者控制台

開發者 Discord

支援

Консоль разработчика

Дискорд для разработчиков

Поддержка

Konsol Pengembang

Discord Pengembang

Dukungan

Console do Desenvolvedor

Discord para Desenvolvedores

Suporte

Anthropic

User Guides

API Reference

Prompt Library

Guides d'utilisation

Référence de l'API

Bibliothèque de prompts

Benutzerhandbücher

API-Referenz

Prompt-Bibliothek

Guide per l'utente

Riferimento API

Libreria di Prompt

ユーザーガイド

API リファレンス

プロンプトライブラリ

Guías de Usuario

Referencia de la API

Biblioteca de Prompts

사용자 가이드

API 참조

프롬프트 라이브러리

用户指南

API 参考

提示库

使用者指南

API 參考文件

提示語料庫

Руководства пользователя

Справочник по API

Библиотека запросов

Panduan Pengguna

Referensi API

Pustaka Prompt

Guias do Usuário

Referência da API

Talk to Claude

Research

News

Overview

Prompt engineering

Welcome to Claude

Quickstart guide

Getting access to Claude

Claude is trained to be a helpful, honest, and harmless assistant. It is used to speaking in _dialogue_, and you can instruct it in regular natural language requests as if you were making requests of a human.

Intro to prompting

Migrating from other models

Upgrading to the Messages API

Models overview

While we highly recommend upgrading to the Claude 3 family of models for their superior performance, vision capabilities, and overall intelligence (see the [models overview](/en/docs/models-overview) page for more information), we understand that transitioning takes time, and some users may need to continue working with our legacy models for a while longer. 

Legacy model guide

These concepts are not unique to Anthropic’s language models, but we present a brief summary of key terms below.

Glossary

Text generation

System prompts

Vision

Embeddings

You can call Claude in Google Sheets with the [Claude for Sheets extension](https://workspace.google.com/marketplace/app/claude%5Ffor%5Fsheets/909417792257). Claude for Sheets enables seamless integration between Claude and Google Sheets, allowing you to execute interactions with Claude directly in cells. This tool allows for easy and rapid prompt engineering by enabling you to construct an evaluation suite and then test different prompts on every item of the evaluation suite in parallel. Separately, we have found Claude for Sheets to be excellent for a variety of office tasks such as processing and categorizing survey results, as well as analyzing tabular data found online.

Google Sheets add-on

Tool use (function calling)

Tool use examples

Tool use requests are priced the same as any other Claude API request, based on the total number of input tokens sent to the model (including in the `tools` parameter) and the number of output tokens generated.

Tool use pricing and tokens

Legacy tool use

Prompt Generator

When interacting with Claude, providing clear and direct instructions is essential for achieving the best results. Think of Claude like a smart but new employee who has no context on what to do aside from what you explicitly tell them. Just as when you instruct a human for the first time on a task, the more you explain exactly what you want in a straightforward manner, the better and more accurate Claude's response will be.

Be clear & direct

Use examples

Claude is a highly capable AI assistant, but sometimes it benefits from having additional context to understand the role it should play in a given conversation. By assigning a role to Claude, you can prime it to respond in a specific way, improve its accuracy and performance, and tailor its tone and demeanor to match the desired context. This technique is also known as role prompting.

Give Claude a role

Use XML tags

You can think of working with large language models like juggling. The more tasks you have Claude handle in a single prompt, the more liable it is to drop something or perform any single task less well. Thus, for complex tasks that require multiple steps or subtasks, we recommend breaking those tasks down into subtasks and chaining prompts to ensure highest quality performance at every step.

Chain prompts

When faced with a complex question or task, it's often beneficial to let Claude think through the problem step-by-step before providing a final answer. This technique, also known as chain of thought (CoT) prompting, can significantly improve the accuracy and nuance of Claude's responses.

Let Claude think

When using Claude, you have the unique ability to guide its responses by prefilling the `Assistant` message. This powerful technique allows you to direct Claude's actions, control the output format, and even help Claude stay in character during role-play scenarios. In some cases where Claude is not performing as expected, a few prefilled sentences can vastly improve Claude's performance.

Prefill Claude's response

Claude is highly capable of producing output in a wide variety of formats. By providing [clear instructions](/en/docs/be-clear-direct), [examples](/en/docs/use-examples), and [prefilled responses](/en/docs/prefill-claudes-response), you can guide Claude to generate responses that adhere to your desired structure and style.

Control output format (JSON mode)

Sometimes, even with a well-crafted prompt, Claude's response might not quite hit the mark or be completely accurate. In these cases, you can leverage Claude's ability to revise its own work by asking for a rewrite. By providing clear instructions and a rubric for the desired output, you can guide Claude to generate content that better aligns with your goals.

Ask Claude for rewrites

Claude's extended context window (200K tokens for Claude 3 models) enables it to handle complex tasks that require processing large amounts of information.

Long context window tips

Sometimes, the hardest part of using an AI model is figuring out how to prompt it effectively.

Helper metaprompt (experimental)

Reducing latency

Large language models ([LLMs](/en/docs/glossary#llm)) like Claude are powerful tools that can generate human-like text based on the input provided. However, due to their generative nature and the vast amount of data they are trained on, users may encounter certain issues while working with these models. This section provides an overview of common problems and offers strategies to mitigate them, ensuring a better user experience and more reliable output.

Troubleshooting

While Claude is incredibly powerful and versatile, it can sometimes generate text that is factually incorrect, inconsistent, or irrelevant to the given context.

Minimizing hallucinations

Mitigating jailbreaks & prompt injections

Due to the generative nature of large language models ([LLMs](/en/docs/glossary#llm)) like Claude, there is a risk that LLMs may reveal parts of the input prompt in its generated output.

Reducing prompt leaks

Sometimes, particularly over long conversations or with legacy Claude models, Claude may slip out of character in character role-playing scenarios.

Keep Claude in character

Use cases and capabilities

Empirical performance evaluations

Screening user input before it reaches your main language model allows you to prevent the processing or output of harmful, offensive, or irrelevant content, saving both computational resources and potential damage to your brand reputation.

Content moderation

Claude excels at processing, understanding, and recognizing patterns in text, images, and data. These capabilities make Claude especially powerful for classification tasks.

Classification

Getting started

Anthropic services live at a fixed range of IP addresses. You can add these to your firewall to open the minimum amount of surface area for egress traffic when accessing the Anthropic API and Console. These ranges will not change without notice.

IP addresses

When making API requests, you must send an `anthropic-version` request header. For example, `anthropic-version: 2023-06-01`. If you are using our [client libraries](/en/api/client-libraries), this is handled for you automatically.

Versions

Errors

To mitigate against misuse and manage capacity on our API, we have implemented limits on how much an organization can use the Claude API.

Rate limits

We provide libraries in Python and Typescript that make it easier to work with the Anthropic API.

Client SDKs

Here are the countries, regions, and territories we can currently support access from:

Supported regions

We've tried to provide the answers to the most common questions in these docs. However, if you need further technical support using Claude, the Anthropic API, or any of our products, you may reach our support team at [support.anthropic.com](https://support.anthropic.com).

Getting help

Create a Message.

Send a structured list of input messages with text and/or image content, and the model will generate the next message in the conversation.

The Messages API can be used for for either single queries or stateless multi-turn conversations.

Create a Message

Streaming Messages

Migrating from Text Completions to Messages

Migrating from Text Completions

Request and response examples for the Messages API

Messages examples

[Legacy] Create a Text Completion.

The Text Completions API is a legacy API. We recommend using the [Messages API](https://docs.anthropic.com/en/api/messages) going forward.

Future models and features will not be compatible with Text Completions. See our [migration guide](https://docs.anthropic.com/en/api/migrating-from-text-completions-to-messages) for guidance in migrating from Text Completions to Messages.

Create a Text Completion

Streaming Text Completions

Prompt validation

Anthropic’s Claude models are now generally available through Amazon Bedrock.

Amazon Bedrock API

Anthropic’s Claude models are now generally available through [Vertex AI](https://cloud.google.com/vertex-ai).

Vertex AI API

Library

Generate an interactive speed typing game in a single HTML file, featuring side-scrolling gameplay and Tailwind CSS styling.

Cosmic Keystrokes

Extract insights, identify risks, and distill key information from long corporate reports into a single memo.

Corporate clairvoyant

Create one-page websites based on user specifications.

Website wizard

Create Excel formulas based on user-described calculations or data manipulations.

Excel formula expert

Generate Google Apps scripts to complete tasks based on user requirements.

Google apps scripter

Python bug buster

Help the user navigate hypothetical time travel scenarios and their implications.

Time travel consultant

Collaboratively create engaging stories with the user, offering plot twists and character development.

Storytelling sidekick

Get answers to questions about a document's content with relevant citations supporting the response.

Cite your sources

Transform everyday language into SQL queries.

SQL sorcerer

Offer interpretations and insights into the symbolism of the user's dreams.

Dream interpreter

Generate clever puns and wordplay based on any given topic.

Pun-dit

Suggest recipe ideas based on the user's available ingredients and dietary preferences.

Culinary creator

Blend two words together to create a new, meaningful portmanteau.

Portmanteau poet

Chat with a knowledgeable AI that has a sarcastic side.

Hal the humorous helper

Write LaTeX documents, generating code for mathematical equations, tables, and more.