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

# Continue

[Continue](https://continue.dev) is an open-source AI coding assistant, available as a VS Code / JetBrains plugin and as a command-line tool (`cn`). Both share the same `config.yaml`, so the configuration below applies to either.

## Configure the CLI

### 1. Install the CLI

<CodeGroup>
  ```bash npm theme={null}
  node -v   # needs 20+ — upgrade via https://nodejs.org/en/download if older
  npm install -g @continuedev/cli
  ```

  ```bash install script (macOS/Linux) theme={null}
  curl -fsSL https://raw.githubusercontent.com/continuedev/continue/main/extensions/cli/scripts/install.sh | bash
  ```
</CodeGroup>

### 2. Configure the model

Create an API Key on the [API Keys](https://console.scitix.ai/model-inference/api_keys) page, then edit `~/.continue/config.yaml`. On a fresh machine, create the directory first with `mkdir -p ~/.continue`.

```yaml theme={null}
name: scitix
version: 0.0.1
schema: v1

models:
  - name: glm-5.2
    provider: openai
    model: glm-5.2
    apiBase: https://api.scitix.ai/model-api/v1
    apiKey: <Your API Key>
    roles:
      - chat
      - edit
      - apply
```

### 3. Create the onboarding marker

As of CLI 1.5.47, interactive `cn` otherwise shows a first-run login prompt asking for an Anthropic API key and ignores your local config. Headless `cn -p` is unaffected. See [continuedev/continue#12258](https://github.com/continuedev/continue/issues/12258).

```bash theme={null}
touch ~/.continue/.onboarding_complete
```

### 4. Run the CLI

Plain `cn` opens the interactive chat TUI. Type `/` to see slash commands. `cn -p "<prompt>"` answers once and exits.

```bash theme={null}
cn                                           # interactive chat
cn -p "introduce yourself in one sentence"  # one-shot
```

For tasks that touch files or run commands, add `--auto` to auto-approve tool calls, or stay in interactive mode to approve each one. Finer-grained flags also exist: `--readonly` for plan mode with read-only tools, and repeatable `--allow <tool>` / `--exclude <tool>`.

## Configure the VS Code / JetBrains Extension

1. Install the extension from the VS Code or JetBrains marketplace by searching for **Continue**.
2. Open the chat sidebar. In VS Code, use `Cmd/Ctrl+L`; in JetBrains, use `Cmd/Ctrl+J`.
3. Click the config selector above the input, then the gear icon next to **Local Config**.
4. Add the same `models` entry shown above to `~/.continue/config.yaml`, or `%USERPROFILE%\.continue\config.yaml` on Windows.
5. Select `glm-5.2` from the model dropdown and send a message.

<Note>
  The CLI and extension share `~/.continue/config.yaml`. If you are logged into Continue Hub, make sure the active config is the local one.
</Note>

## Field Reference

The CLI and the extension share `config.yaml`, so the same fields apply to both.

| Field                      | Description                                                                                                                                                        |
| -------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| `provider`                 | Keep as `openai` for any OpenAI-compatible service.                                                                                                                |
| `model`                    | Model ID, such as `glm-5.2`.                                                                                                                                       |
| `apiBase`                  | ScitiX OpenAI-compatible endpoint: `https://api.scitix.ai/model-api/v1`.                                                                                           |
| `apiKey`                   | Your API Key from the [API Keys](https://console.scitix.ai/model-inference/api_keys) page.                                                                         |
| `roles`                    | Continue features that use this model. Valid values: `chat`, `autocomplete`, `embed`, `rerank`, `edit`, `apply`.                                                   |
| `capabilities`             | Optional. Set `[tool_use]` if agent tool calls do not work out of the box. This overrides Continue's autodetection, which can miss on OpenAI-compatible endpoints. |
| `defaultCompletionOptions` | Optional. Set `contextLength` / `maxTokens` to the model's actual limits.                                                                                          |

<Tip>
  For Tab autocomplete, add a second lightweight model entry with `roles: [autocomplete]`. A fast model such as `DeepSeek-V4-Flash` works well.
</Tip>

## FAQ

* The chat TUI shows a **Model Capability Warning** for `glm-5.2`. As of 1.5.47, the CLI checks model IDs against a built-in list and warns on ones it does not recognize, even with `capabilities` declared. It is safe to ignore: chat, edit, and tool calls all work against the ScitiX endpoint.