Skip to main content

VS Code with Continue and Verdictan

Continue supports an apiBase override for OpenAI-compatible models. Use its config.yaml format. Continue has deprecated the older config.json format.

Store the client token​

Continue's IDE extension resolves local secrets from a workspace .env, .continue/.env, or ~/.continue/.env. Put the client token in one of those git-ignored files:

VERDICTAN_IDE_TOKEN=<client-api-token>

After you change the secret, start VS Code again.

Configure the model​

Add a model to Continue's config.yaml:

name: Verdictan
version: 1.0.0
schema: v1

models:
  - name: Verdictan Chat
    provider: openai
    model: your-model-id
    apiBase: http://127.0.0.1:41002/v1
    apiKey: ${{ secrets.VERDICTAN_IDE_TOKEN }}
    roles:
      - chat
      - edit
      - apply

Replace your-model-id with a model returned by the gateway's /v1/models endpoint. If the upstream model and Continue configuration do not support fill-in-the-middle completion, do not assign the autocomplete role.

See Continue's config reference for the model fields and Continue's secret guidance for the resolution sequence.

Verify​

Run verdictan events tail --since 10m --follow. Select Verdictan Chat in Continue. Send a request. If Continue continues to call a different provider, check the active configuration and selected model. Do not use VS Code's generic proxy setting.

Next steps​