Continue CLI: config.yaml and API Setup

Continue CLI installs as the cn command and reads the same YAML configuration schema used by Continue in an editor. An OpenAI-compatible setup needs provider: openai, a Base URL, a current Model ID, and a safe API-key reference. Verify installation with cn --version, then run one short interactive or headless request.

Want to test Continue CLI without storing a secret in your repository? Create a BetterToken account and keep the current Continue CLI guide open. The Model ID remains a placeholder because it is dynamic.

Prerequisites

You need a terminal that can access the official installer, your own BetterToken API key, a current Model ID, a user directory for local configuration, and a harmless test prompt.

BetterToken covers the OpenAI-compatible API endpoint. It does not guarantee access to GitHub, npm, the Continue website, installers, or updates. The BetterToken endpoint can be reached from Russia without a VPN; other network steps remain separate.

1. Install Continue CLI

The current Continue documentation provides a shell installer for macOS/Linux:

curl -fsSL https://raw.githubusercontent.com/continuedev/continue/main/extensions/cli/scripts/install.sh | bash

Use the current official PowerShell command on Windows. If you prefer npm and have a supported Node.js version:

npm install -g @continuedev/cli

Do not run both installers after the first error without checking which binary and directory already exist. Verify the result:

cn --version

2. Identify the active config

Continue CLI resolves configuration in this order: explicit --config <path>, the last persisted config, and the default user file. The default path is ~/.continue/config.yaml on macOS/Linux and %USERPROFILE%\.continue\config.yaml on Windows.

If a change is ignored, test with an explicit absolute config path instead of editing several copies.

3. Add an OpenAI-compatible model

name: BetterToken Continue CLI
version: 0.0.1
schema: v1

models:
  - name: BetterToken model
    provider: openai
    model: YOUR_MODEL_ID
    apiBase: https://www.bettertoken.ai/v1
    apiKey: ${{ secrets.BETTERTOKEN_API_KEY }}

Replace the Model ID with the current exact value. Confirm the current secret-reference syntax in Continue documentation. If the CLI uses a process environment variable, set it in the local session. Never commit a real key.

4. Test interactive and headless modes

Start with an explicit config:

cn --config ~/.continue/config.yaml

Ask the tool to reply with CONTINUE_OK without changing files. For a one-shot check, use the current documented cn -p form:

cn --config ~/.continue/config.yaml -p "Reply with CONTINUE_OK and do not change files."

The test passes when the command returns normally, uses the expected Model ID, and creates a matching BetterToken Dashboard record with status and token usage.

Common failures

For command not found, restart the shell and inspect PATH before reinstalling. If the CLI ignores YAML, pass --config explicitly. For 401, verify the secret reference and its availability in the current process without printing the value. For 404, use the Base URL https://www.bettertoken.ai/v1 without an appended resource path. For model not found, copy the current ID from the model catalog.

A short request confirms that cn read the intended YAML, obtained the secret, reached the endpoint, and used an available model. It does not verify every cloud feature, tool, or remote-session integration.

Related articles