Use the Anthropic adapter to connect to Claude 3.5 Sonnet, Claude 3 Opus and more

The Anthropic adapter provides complete integration with the Anthropic Claude API, supporting the Messages API, tool use, vision capabilities, and streaming.

Installation

pnpm add @amux.ai/llm-bridge @amux.ai/adapter-anthropic

Basic Usage

import { createBridge } from '@amux.ai/llm-bridge'
import { anthropicAdapter } from '@amux.ai/adapter-anthropic'

const bridge = createBridge({
  inbound: anthropicAdapter,
  outbound: anthropicAdapter,
  config: {
    apiKey: process.env.ANTHROPIC_API_KEY
  }
})

const response = await bridge.chat({
  model: 'claude-3-5-sonnet-20241022',
  max_tokens: 1024,
  messages: [
    { role: 'user', content: 'What is Amux?' }
  ]
})

console.log(response.content[0].text)

The Anthropic API requires the max_tokens parameter to be specified.

Supported Features

Tool Use

const response = await bridge.chat({
  model: 'claude-3-5-sonnet-20241022',
  max_tokens: 1024,
  messages: [
    { role: 'user', content: 'What time is it in Beijing?' }
  ],
  tools: [{
    name: 'get_current_time',
    description: 'Get the current time for a specified city',
    input_schema: {
      type: 'object',
      properties: {
        city: {
          type: 'string',
          description: 'City name'
        }
      },
      required: ['city']
    }
  }]
})

// Check for tool calls
if (response.stop_reason === 'tool_use') {
  const toolUse = response.content.find(c => c.type === 'tool_use')
  console.log('Tool:', toolUse.name)
  console.log('Input:', toolUse.input)
}

Vision Capabilities

const response = await bridge.chat({
  model: 'claude-3-5-sonnet-20241022',
  max_tokens: 1024,
  messages: [{
    role: 'user',
    content: [
      {
        type: 'text',
        text: 'What is in this image?'
      },
      {
        type: 'image',
        source: {
          type: 'url',
          url: 'https://example.com/image.jpg'
        }
      }
    ]
  }]
})

Supported Image Formats:

URL (https://)
Base64 encoded (requires media_type)

System Prompt

Claude uses a separate system parameter:

const response = await bridge.chat({
  model: 'claude-3-5-sonnet-20241022',
  max_tokens: 1024,
  system: 'You are a helpful AI assistant specializing in technical questions.',
  messages: [
    { role: 'user', content: 'What is TypeScript?' }
  ]
})

Streaming

const stream = bridge.chatStream({
  model: 'claude-3-5-sonnet-20241022',
  max_tokens: 1024,
  messages: [
    { role: 'user', content: 'Tell me a story' }
  ],
  stream: true
})

for await (const event of stream) {
  if (event.type === 'content_block_delta') {
    if (event.delta.type === 'text_delta') {
      process.stdout.write(event.delta.text)
    }
  }
}

Extended Thinking (Reasoning)

Claude supports extended thinking for complex reasoning tasks:

const response = await bridge.chat({
  model: 'claude-3-7-sonnet-20250219',
  max_tokens: 4096,
  messages: [
    { role: 'user', content: 'Solve this complex problem step by step...' }
  ],
  thinking: {
    type: 'enabled',
    budget_tokens: 2000
  }
})

// Access reasoning content
for (const block of response.content) {
  if (block.type === 'thinking') {
    console.log('Reasoning:', block.thinking)
  } else if (block.type === 'text') {
    console.log('Answer:', block.text)
  }
}

Supported Models

Model	Context Length	Description
`claude-3-5-sonnet-20241022`	200K	Latest and most capable
`claude-3-opus-20240229`	200K	Most intelligent model
`claude-3-sonnet-20240229`	200K	Balanced performance and speed
`claude-3-haiku-20240307`	200K	Fastest and most economical

Configuration Options

const bridge = createBridge({
  inbound: anthropicAdapter,
  outbound: anthropicAdapter,
  config: {
    apiKey: process.env.ANTHROPIC_API_KEY,
    baseURL: 'https://api.anthropic.com', // Optional
    timeout: 60000, // Optional
    headers: {
      'anthropic-version': '2023-06-01' // API version
    }
  }
})

Converting with OpenAI Format

OpenAI → Anthropic

import { openaiAdapter } from '@amux.ai/adapter-openai'
import { anthropicAdapter } from '@amux.ai/adapter-anthropic'

const bridge = createBridge({
  inbound: openaiAdapter,
  outbound: anthropicAdapter,
  config: {
    apiKey: process.env.ANTHROPIC_API_KEY
  }
})

// Send request in OpenAI format
const response = await bridge.chat({
  model: 'gpt-4',
  messages: [{ role: 'user', content: 'Hello' }]
})
// Returns OpenAI format response

Anthropic → OpenAI

const bridge = createBridge({
  inbound: anthropicAdapter,
  outbound: openaiAdapter,
  config: {
    apiKey: process.env.OPENAI_API_KEY
  }
})

// Send request in Anthropic format
const response = await bridge.chat({
  model: 'claude-3-5-sonnet-20241022',
  max_tokens: 1024,
  messages: [{ role: 'user', content: 'Hello' }]
})
// Returns Anthropic format response

Feature Support

Feature	Supported	Notes
Messages API	✅	Fully supported
Streaming	✅	Fully supported
Tool Use	✅	Fully supported
Vision	✅	Claude 3 series
System Prompt	✅	Separate system parameter
Long Context	✅	200K tokens
Extended Thinking	✅	Reasoning support

Best Practices

1. Always Set max_tokens

// ✅ Correct
const response = await bridge.chat({
  model: 'claude-3-5-sonnet-20241022',
  max_tokens: 1024, // Required
  messages: [...]
})

// ❌ Wrong - will error
const response = await bridge.chat({
  model: 'claude-3-5-sonnet-20241022',
  messages: [...] // Missing max_tokens
})

2. Use System Prompts to Optimize Responses

const response = await bridge.chat({
  model: 'claude-3-5-sonnet-20241022',
  max_tokens: 1024,
  system: 'You are a professional technical documentation writer. Use clear, concise language.',
  messages: [
    { role: 'user', content: 'Explain what a REST API is' }
  ]
})

3. Handle Long Conversations

// Claude supports 200K tokens context
const response = await bridge.chat({
  model: 'claude-3-5-sonnet-20241022',
  max_tokens: 2048,
  messages: [
    { role: 'user', content: 'First question' },
    { role: 'assistant', content: 'First answer' },
    { role: 'user', content: 'Second question' },
    { role: 'assistant', content: 'Second answer' },
    // ... can have many turns
    { role: 'user', content: 'Latest question' }
  ]
})

Next Steps

OpenAI Adapter

Learn how to use GPT models

DeepSeek Adapter

Learn how to use DeepSeek models

Adapter API

View the complete Adapter API reference

Anthropic Adapter