Ir al contenido
Nexus Docs
↑↓Navigate↵SelectescClose

Continue (VS Code) → Nexus

Route the Continue VS Code extension through the Nexus OpenAI-compatible edge

Last updated on:

Esta página aún no está disponible en tu idioma.

Route the Continue VS Code extension through the Nexus OpenAI-compatible edge (/v1). Continue’s openai provider accepts a custom apiBase; Nexus verifies a gateway API key (nxs_…) and dispatches to the configured upstream.

Shared prerequisites (gateway URL, nxs_… key, registry model, BYOK/PTB): Client Integrations.

1. Open local config

Section titled “1. Open local config”

Continue loads ~/.continue/config.yaml (Global / Local Config in the assistant picker). Ensure the file has top-level name, version, and schema so the extension can parse it.

2. Add Nexus models

Section titled “2. Add Nexus models”
name: Nexus Continue
version: 0.0.1
schema: v1


models:
    - name: Nexus chat
      provider: openai
      model: openai/gpt-5.4-mini
      apiBase: https://api.example.com/v1
      apiKey: nxs_your_prefix_your_secret
      roles:
          - chat
          - edit
          - apply
      capabilities:
          - tool_use


    - name: Nexus autocomplete
      provider: openai
      model: openai/gpt-5.4-mini
      apiBase: https://api.example.com/v1
      apiKey: nxs_your_prefix_your_secret
      roles:
          - autocomplete

Local Compose: set apiBase to http://127.0.0.1:14450/v1.

Notes:

  • apiBase is the version root (…/v1), not …/v1/chat/completions.
  • model must be an id returned by GET /v1/models.
  • Prefer Continue’s secret / env mechanisms over committing raw keys when your Continue version supports them.
  • Some Continue builds prefer the Responses API for certain GPT / reasoning models. Nexus supports POST /v1/responses for OpenAI-routed models. If a model fails only inside Continue and the curl chat check below succeeds, set useResponsesApi: false on that model entry to force chat completions:
- name: Nexus chat (completions)
  provider: openai
  model: openai/gpt-5.4-mini
  apiBase: https://api.example.com/v1
  apiKey: nxs_your_prefix_your_secret
  useResponsesApi: false
  roles:
      - chat
      - edit
      - apply

3. Select Local Config and smoke-test

Section titled “3. Select Local Config and smoke-test”
  1. In Continue, choose the Local Config assistant (not a Hub assistant that ignores config.yaml).
  2. Pick Nexus chat and send a short prompt.
  3. Confirm Chat / Edit / Apply as needed.

Independent check:

Terminal window
GATEWAY_URL="${GATEWAY_URL:-https://api.example.com}"
NEXUS_API_KEY='nxs_…'


curl -sS "$GATEWAY_URL/v1/chat/completions" \
  -H "Authorization: Bearer $NEXUS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/gpt-5.4-mini",
    "messages": [{"role": "user", "content": "ping"}],
    "max_tokens": 16
  }' | jq .

Common failures

Section titled “Common failures”
Symptom Likely cause
No models / config error Missing top-level name / version / schema
Models missing in UI Hub assistant selected instead of Local Config
401 nxd_… or upstream key used instead of nxs_…
Wrong path / 404 apiBase missing /v1 or includes /chat/completions
Responses-shaped failures Try useResponsesApi: false, or use an OpenAI-routed model that supports /v1/responses
Section titled “Related”