Table of Contents
Do not index
In this walkthrough, you’ll set up OpenClaw with a Claude API key from the Agent37 dashboard. By the end, you’ll have a running OpenClaw instance, Anthropic set as the provider, and a quick way to verify everything is working.
Follow the steps below or watch the video.
Video walkthrough

Prerequisites
- An active Agent 37 account
- A running OpenClaw instance inside Agent 37
- An Anthropic Console account
- A valid Claude (Anthropic) API key
Step 1: Create your Claude (Anthropic) API key
- Go to the Claude Console and sign in. (Anthropic's console moved from console.anthropic.com, which now redirects here.)
- Open Settings > API keys and click Create key. Copy the generated key, the Console shows it only once.
- Click
Create Key.
- When prompted:
- Select the workspace.
- Enter a name for the key.
- Click Add.
- Copy the generated key.


Claude API keys typically look like:
sk-ant-xxxx...Step 2: Open your Agent 37 instance
- Go to the Agent37 dashboard and sign in.
- Open an existing instance (or create a new one).
- Open Terminal.
Step 3: Configure OpenClaw to use Anthropic
In Terminal, run:
openclaw configureWhen prompted, choose the following:
- Where will the gateway run? >
Local

Continue with:
- What do you want to configure? > Model

- Model / Auth Provider >
Anthropic

- Anthropic Auth Model >
Anthropic API Key

- How do you want to provide this API key now? >
Paste API key now

- Enter Anthropic API key > paste the key you copied in Step 1

- Models in /model picker > select your model (example:
anthropic/claude-sonnet-5; on OpenClaw builds from before July 2026, useanthropic/claude-sonnet-4-6)

- After selecting the model >
Continue

Step 4: Verify the configured model
Run:
openclaw models listYou should see the model you selected set as the default in the output.

You have successfully configured the Claude API key.
If you are not sure which model to choose, start with a Claude Sonnet model. It is usually a strong default for general OpenClaw workflows, though our full model ranking shows when Sonnet is not the right call.
What to do next
- Compare with the OpenAI API key setup if you also use OpenAI models.