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

# Continue

> Use derestricted.ai for chat, edits, and tool use in Continue.

Configure Continue's OpenAI-compatible provider with your derestricted key and model.

## Before you start

[Install the Continue extension](https://docs.continue.dev/ide-extensions/install) and create a key in the [console](https://derestricted.ai/console/). New verified accounts receive \$5 in starting credit.

## Store your key locally

Add this line to `~/.continue/.env`, replacing the placeholder:

```dotenv theme={"system"}
DERESTRICTED_API_KEY=glmk_YOUR_API_KEY
```

Continue's IDE extension resolves local secrets from `.env` files; exporting a key in a terminal is not sufficient. Keep this file private and restart the editor after changing it. See Continue's [secret configuration](https://docs.continue.dev/faqs#managing-local-secrets-and-environment-variables).

## Add the model

Open your local Continue configuration, normally `~/.continue/config.yaml`. Merge the model entry below into its `models` list. If you are creating the file, use the full example:

```yaml theme={"system"}
name: derestricted.ai
version: 1.0.0
schema: v1
models:
  - name: derestricted-llm
    provider: openai
    model: derestricted-llm
    apiBase: https://api.derestricted.ai/v1
    apiKey: ${{ secrets.DERESTRICTED_API_KEY }}
    roles:
      - chat
      - edit
      - apply
    capabilities:
      - tool_use
    defaultCompletionOptions:
      maxTokens: 8192
```

`apiBase` selects the compatible endpoint. The explicit `tool_use` capability lets Continue offer Agent mode for a custom model. See the official [OpenAI-compatible setup](https://docs.continue.dev/customize/model-providers/top-level/openai) and [configuration reference](https://docs.continue.dev/reference#models).

## Start using it

Select your local configuration in Continue, choose `derestricted-llm`, and ask it to explain a selected function. Use Chat for questions, Edit for changes, or Agent mode for tasks involving tools.

## If something fails

* **Secret not found:** check `~/.continue/.env` and restart the editor. Workspace `.env` files can override the global value.
* **Agent mode unavailable:** ensure `capabilities` includes `tool_use` and the selected model is `derestricted-llm`.
* **Other model roles:** this example configures chat and edits. The gateway has no embeddings or legacy `/completions` endpoint. Use another provider for embeddings and for any autocomplete model that requires `/completions`.
* **Credit, size, or rate-limit errors:** see [API compatibility](/api/introduction) and [troubleshooting](/api/errors).