> ## Documentation Index
> Fetch the complete documentation index at: https://docs.getcargo.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Anthropic

> This page outlines how to use the Anthropic integration for AI-powered actions in Cargo.

Anthropic builds reliable, interpretable AI systems. Claude, their AI assistant, excels at thoughtful dialogue, content creation, complex reasoning, and coding. Cargo's native integration with Anthropic allows you to use Claude models directly in your workflows.

## How to set up Anthropic

You can use Anthropic in two ways:

1. **Cargo credits** – Use Claude through Cargo's managed integration
2. **Your own API key** – Connect your Anthropic account

### Connection details

| Field       | Description            |
| ----------- | ---------------------- |
| **API Key** | Your Anthropic API key |

<Tip>Find your API key in the Anthropic Console under **API Keys**.</Tip>

***

## Anthropic actions

### Instruct

Send a prompt to a Claude model and receive a text response.

**Use cases**

* **Content generation** – Generate personalized emails and messages
* **Classification** – Categorize text into predefined categories
* **Summarization** – Summarize long documents or conversations
* **Data extraction** – Extract structured information from text
* **Analysis** – Analyze sentiment, intent, or other properties
* **Reasoning** – Complex multi-step problem solving

**Configuration**

| Field               | Description                     |
| ------------------- | ------------------------------- |
| **Model**           | Select the Claude model to use  |
| **Prompt**          | The prompt to send to the model |
| **System prompt**   | Optional system instructions    |
| **Maximum tokens**  | Limit output length             |
| **Temperature**     | Control randomness (0-1)        |
| **With web search** | Enable web search capabilities  |

### Available models

The available Claude models — and their per-model credit costs — evolve with Anthropic's lineup, so the action's **Model** dropdown in the workspace is always the authoritative list. From the terminal:

```bash theme={null}
cargo-ai connection integration get anthropic
```

**Picking a model:**

* **Opus** — the most powerful tier: complex reasoning, advanced coding, creative work (highest credit cost)
* **Sonnet** — the best balance of intelligence and cost for most tasks, including agents (start here)
* **Haiku** — fast, low-cost inference for high-volume, low-latency work

### Advanced settings

#### System prompt

Set instructions that guide Claude's behavior:

* Define the model's role or persona
* Set output format guidelines
* Provide context about your use case
* Establish constraints or rules

#### Temperature

Control the randomness of outputs:

| Temperature | Behavior              |
| ----------- | --------------------- |
| **0**       | Most deterministic    |
| **0.3-0.5** | Balanced              |
| **0.7-1**   | More creative, varied |

#### Web search

Enable Claude to search the web for current information:

* Real-time data lookups
* Current events research
* Fact verification