XylemNode

Integration guides

Continue

An open-source AI coding assistant for VS Code and JetBrains — chat, autocomplete and edit in one. It manages models with a single config.yaml; set provider to openai and change apiBase to connect XylemNode.

Get API Key
Site / download
  1. 01

    Install the extension

    In VS Code, press Ctrl/Cmd+Shift+X, search for Continue and install; JetBrains users search Continue in the Plugins marketplace. After installing, a Continue icon appears in the sidebar — open it for the panel.

  2. 02

    Open config.yaml

    Open the Continue sidebar and click the gear ⚙️ in the top-right to open config.yaml; you can also edit the file directly in your home directory. Newer versions use YAML — the old config.json is deprecated.

    macOS / Linux
    ~/.continue/config.yaml
    Windows
    %USERPROFILE%\.continue\config.yaml
  3. 03

    Add the XylemNode model

    Add an entry to the models list in config.yaml: set provider to openai and point apiBase at XylemNode with /v1. The name / version / schema at the top of the file are auto-generated by Continue — leave them as is.

    models:
  - name: XylemNode
    provider: openai
    model: gpt-5.6-sol
    apiBase: https://api.xylemnode.com/v1
    apiKey: sk-your-key
  4. 04

    Select the model and start

    After saving, return to the Continue panel and select XylemNode from the model dropdown to chat and edit. To change models later, edit only the model line — apiBase and apiKey stay untouched.

Note for gpt-5 series models

For o-series and gpt-5-series model names, Continue defaults to OpenAI's /responses endpoint. If your chosen model (e.g. gpt-5.6-sol) uses /chat/completions on the XylemNode side and thus errors, add useResponsesApi: false under that model to force /chat/completions.

Common questions

The model answers in Continue, but tool calls or image input don't work

Continue infers what a model supports from the provider and the model name. Its own documentation notes that this detection works well for popular models but may not cover custom deployments or newer ones — and a custom endpoint is exactly the case where it may not recognise the name. Declare the capabilities explicitly in config.yaml. Mind one limit: declaring them can only **add** capabilities, it cannot override what Continue already inferred. Also make sure you are editing config.yaml; the older config.json is marked deprecated.