Continue CLI Integration

Continue CLI runs a coding assistant in your terminal. Connect its OpenAI-compatible provider to LLM Gateway with a pay-as-you-go or DevPass API key.

Video walkthrough

Install

pnpm add -g @continuedev/cli
cn --version

Configure the model

Create an API key in your dashboard. Save this configuration in ~/.continue/config.yaml, merging it with any settings you already use:

name: LLM Gateway
version: 1.0.0
schema: v1
models:
  - name: Gateway coding model
    provider: openai
    model: MODEL_ID
    apiBase: https://api.llmgateway.io/v1
    apiKey: ${{ secrets.LLMGATEWAY_API_KEY }}
    capabilities:
      - tool_use

Replace MODEL_ID with a text model that supports tools from the live catalogue, and export your key in the terminal where you launch Continue:

export LLMGATEWAY_API_KEY="your_api_key"

Keep the key private. capabilities: [tool_use] tells Continue to enable tool calling; the selected model must actually support it.

provider: openai selects the API format. LLM Gateway routes the request to the serving provider. With DevPass, use a canonical model ID supported by your plan, without an upstream provider prefix.

Start a session

Launch Continue from your project with the configuration selected explicitly:

cd your-project
cn --config ~/.continue/config.yaml

Ask it to fix a small failing test, leaving the test itself unchanged. Review requested tool permissions, inspect the diff, and confirm the test output. Requests appear in the dashboard for the workspace that issued your key.

For a one-shot prompt:

cn --config ~/.continue/config.yaml -p "Explain this project without changing files"

Change models

Add another entry under models with a different name and model ID, keeping the gateway endpoint. Select the configured model in Continue. Check the live catalogue for each model's tool and image capabilities.

A separate project configuration works too: pass its path to --config when launching the CLI.

Troubleshooting

• Configuration rejected: Include name, version, and schema: v1 at the top level, and check YAML indentation.
• Onboarding appears instead: Pass the intended file with --config.
• Authentication failed: Check the key and workspace in the dashboard.
• Request finishes without visible text: Continue CLI 1.5.47 drops text chunks that also contain usage, which can affect Gemini responses through the gateway. Choose another tool-capable model and check for a Continue update.
• Tool calling unavailable: Choose a model with tool support and include tool_use in its capabilities.
• Connection failed: Use https://api.llmgateway.io/v1 as apiBase.

See the Continue configuration reference for additional options.