prompt-executor-anthropic-client

A client implementation for executing prompts using Anthropic's Claude models with support for images and documents.

Overview

This module provides a client implementation for the Anthropic API, allowing you to execute prompts using Claude models. It handles authentication, request formatting, response parsing, and multimodal content encoding specific to Anthropic's API requirements.

Supported Models

Model	Speed	Context	Input Support	Output Support	Pricing (per 1M tokens)
Claude 3 Opus	Moderately fast	200K	Text, Images, PDF	Text, Tools	$15.00 / $ 75.00
Claude 3 Haiku	Fast	200K	Text, Images, PDF	Text, Tools	$0.25 / $ 1.25
Claude 3.5 Haiku	Fastest	200K	Text, Images, PDF	Text, Tools	$0.8 / $ 4.0
Claude 3.5 Sonnet	Fast	200K	Text, Images, PDF	Text, Tools	$3.00 / $ 15.00
Claude 3.7 Sonnet	Fast	200K	Text, Images, PDF	Text, Tools	$3.00 / $ 15.00
Claude Sonnet 4	Fast	200K	Text, Images, PDF	Text, Tools	$3.00 / $ 15.00
Claude Opus 4	Moderately fast	200K	Text, Images, PDF	Text, Tools	$15.00 / $ 75.00

Media Content Support

Content Type	Supported Formats	Max Size	Notes
Images	PNG, JPEG, WebP, GIF	5MB	Base64 encoded or URL
Documents	PDF (URL/local), TXT, MD	5MB	PDF passed directly, TXT/MD as plain text
Audio	❌ Not supported	-	-
Video	❌ Not supported	-	-

Document Support Details:

PDF: Supported via URL or local file (base64 encoded)
TXT/MD: Read as plain text content
Other formats: Not supported (HTML, CSV, XML, etc.)

Using in your project

Add the dependency to your project:

dependencies {
    implementation("ai.koog.prompt:prompt-executor-anthropic-client:$version")
}

Configure the client with your API key:

val anthropicClient = AnthropicLLMClient(
    apiKey = "your-anthropic-api-key",
)

Using in tests

For testing, you can use a mock implementation:

val mockAnthropicClient = MockAnthropicClient(
    responses = listOf("Mocked response 1", "Mocked response 2")
)

Example of usage

suspend fun main() {
    val client = AnthropicLLMClient(
        apiKey = System.getenv("ANTHROPIC_API_KEY"),
    )

    // Text-only example
    val response = client.execute(
        prompt = prompt {
            system("You are helpful assistant")
            user("What time is it now?")
        },
        model = AnthropicModels.Sonnet_3_5
    )

    println(response)
}

Multimodal Examples

// Image analysis
val imageResponse = client.execute(
    prompt = prompt {
        user {
            text("What do you see in this image?")
            image("/path/to/image.jpg")
        }
    },
    model = AnthropicModels.Sonnet_3_5
)

// PDF document processing
val pdfResponse = client.execute(
    prompt = prompt {
        user {
            text("Summarize this PDF document")
            document("/path/to/document.pdf")
        }
    },
    model = AnthropicModels.Sonnet_3_5
)

// PDF from URL
val pdfUrlResponse = client.execute(
    prompt = prompt {
        user {
            text("Analyze this research paper")
            document("https://example.com/paper.pdf")
        }
    },
    model = AnthropicModels.Sonnet_3_5
)

// Text file processing
val textResponse = client.execute(
    prompt = prompt {
        user {
            text("Review this markdown file")
            document("/path/to/readme.md")
        }
    },
    model = AnthropicModels.Sonnet_3_5
)

// Mixed content
val mixedResponse = client.execute(
    prompt = prompt {
        user {
            text("Compare this image with the PDF:")
            image("/path/to/chart.png")
            document("/path/to/report.pdf")
            text("What insights can you provide?")
        }
    },
    model = AnthropicModels.Sonnet_3_5
)

Packages

ai.koog.prompt.executor.clients.anthropic

common