# API Access & Client Setup

Use your Coding Plan key from the API directly or wire it into Claude Code, OpenClaw, Cursor, Codex, and other coding clients.

Source: https://atlascloud.ai/docs/en/coding-plan/api

## Calling the API Directly

Coding Plan traffic flows through a **dedicated API key** that is separate from your standard Atlas Cloud key.

- **Base URL:** `https://api.atlascloud.ai/v1`
- **API Key:** Generate one from **Plan Management → [Get API Key](https://www.atlascloud.ai/console/api-keys)**.

![Get a Coding Plan API key](../../img/coding-plan-api-key.png)

Request and response shapes follow the OpenAI-compatible specification — see [Model APIs](https://www.atlascloud.ai/docs/models/get-start) for full examples.

## Client Configuration

Most coding tools (OpenClaw, Hermes, OpenCode, Cursor, Copilot, …) take an **OpenAI-compatible** endpoint:

```
https://api.atlascloud.ai/v1
```

Claude Code, which speaks the **Anthropic Messages** protocol, points at:

```
https://api.atlascloud.ai
```

The sections below walk through the most common tools.

### Claude Code

#### Configure environment variables

> **Important:** replace `ANTHROPIC_AUTH_TOKEN` below with the API key you generated on the Atlas Cloud console.

Settings file location:

- **Windows:** `%USERPROFILE%\.claude\settings.json`
- **macOS / Linux:** `~/.claude/settings.json`

Edit `settings.json` and set the API key and the model you want to route through:

```json
{
  "env": {
    "ANTHROPIC_AUTH_TOKEN": "atlas-api-key",
    "ANTHROPIC_BASE_URL": "https://api.atlascloud.ai",
    "ANTHROPIC_MODEL": "zai-org/glm-5.1",
    "ANTHROPIC_DEFAULT_HAIKU_MODEL": "zai-org/glm-5.1",
    "ANTHROPIC_DEFAULT_SONNET_MODEL": "zai-org/glm-5.1",
    "CLAUDE_CODE_DISABLE_EXPERIMENTAL_BETAS": "1"
  }
}
```

### OpenClaw

#### 1. Run the OpenClaw onboarding wizard

After installation, open your terminal (SSH / CMD / Terminal) and launch the wizard:

```bash
openclaw onboard
```

Select **Yes → QuickStart → Custom Provider**, then:

![OpenClaw onboard wizard](../../img/coding-plan-openclaw-onboard.png)

1. Enter the API Base URL: `https://api.atlascloud.ai/v1`
2. Paste your Atlas Cloud API key.
3. Paste the model ID. Choose the **OpenAI-compatible** protocol.
4. Verify the connection — you should see `Verification successful`.
5. Give the endpoint a custom ID and display name.

#### 2. Or configure the settings file directly

- **macOS / Linux:** `~/.claude/settings.json`

```json
{
  "baseUrl": "https://api.atlascloud.ai/v1",
  "apiKey": "your-atlas-key",
  "api": "openai-completions",
  "models": [
    {
      "id": "zai-org/glm-5.1",
      "name": "zai-org/glm-5.1",
      "contextWindow": 200000,
      "input": ["text"]
    }
  ]
}
```

### Hermes

Add a custom OpenAI-compatible provider in Hermes to plug in Atlas Cloud:

1. In the provider list, choose **Custom endpoint (enter URL manually)**.

   ![Hermes — select Custom endpoint](../../img/coding-plan-hermes-provider.png)

2. Set **API base URL** to `https://api.atlascloud.ai/v1` and paste your Coding Plan API Key as the **API key**. Once the endpoint is verified, the available models show up:

   ![Hermes — configure OpenAI-compatible endpoint](../../img/coding-plan-hermes-endpoint.png)

### Trae

In Trae, add a custom OpenAI-compatible provider with the Atlas Cloud base URL and your Coding Plan API key.

![Trae configuration](../../img/coding-plan-trae.png)

### Cursor

In **Cursor → Settings → Models**, add a custom OpenAI-compatible model pointing at `https://api.atlascloud.ai/v1`.

![Cursor configuration](../../img/coding-plan-cursor.png)

### OpenCode

Open the OpenCode config file `opencode.json`:

- **Linux:** `~/.config/opencode/opencode.json`
- **Windows:** `%USERPROFILE%\.config\opencode\opencode.json`

#### Provider config

```json
{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "atlascloud": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "atlascloud",
      "options": {
        "baseURL": "https://api.atlascloud.ai/v1",
        "apiKey": "your-atlas-api-key"
      },
      "models": {
        "zai-org/glm-5.1": {
          "name": "glm-5.1"
        }
      }
    }
  }
}
```

### Codex

#### 1. Register the provider

On macOS or Linux, create `~/.codex/config.toml`:

```toml
model_provider = "atlas_coding_plan"
model = "zai-org/glm-5.1"

[model_providers.atlas_coding_plan]
name = "atlascloud"
base_url = "https://api.atlascloud.ai/v1"
wire_api = "chat"
requires_openai_auth = true
```

#### 2. Add your API key

Create `~/.codex/auth.json`:

```json
{
  "OPENAI_API_KEY": "atlas-api-key"
}
```

> Replace `atlas-api-key` with the key you generated on the Atlas Cloud console.

#### 3. Start Codex

```bash
codex
```

Skip the update prompt and you're ready to go.

![Codex running](../../img/coding-plan-codex.png)