Continue.dev: VSCode power setup

Prereq

VSCode 1.90+ with Continue extension installed
At least two provider keys (diversity matters)
Familiarity with settings.json

Steps

01. Open the config

bash

code ~/.continue/config.yaml

02. Register multiple models

yaml

models:
- name: opus-deep
  provider: anthropic
  model: claude-opus-4.7
  apiKey: ${ANTHROPIC_API_KEY}
  roles: [chat, edit]
- name: gpt-fast
  provider: openai
  model: gpt-5.4-mini
  apiKey: ${OPENAI_API_KEY}
  roles: [autocomplete]
- name: local-qwen
  provider: ollama
  model: qwen3-coder:32b
  roles: [chat, edit]
- name: ovth-router
  provider: openai
  model: auto
  apiBase: https://gateway.ovth.dev/v1
  apiKey: ${OVTH_API_KEY}

03. Bind the three shortcuts that matter

json

// keybindings.json
[
{ "key": "cmd+l",        "command": "continue.focusContinueInput" },
{ "key": "cmd+i",        "command": "continue.quickEdit" },
{ "key": "cmd+shift+l",  "command": "continue.quickEditSelection" }
]

04. Add a custom context provider

yaml

context:
- provider: code
- provider: diff
- provider: terminal
- provider: currentFile
- provider: branch   # custom — pipes `git diff main...HEAD`
- provider: url
  params:
    maxUrls: 2

@branch in chat now feeds the PR-shaped diff to the model. Way sharper than pasting files manually.

05. Switch roles mid-edit

In the chat box: @opus-deep for refactors, @gpt-fast for autocomplete, @local-qwen when offline.

Bridge Continue to OVTH Gateway
Build your first custom slash command

Prereq

Steps

01. Open the config

02. Register multiple models

03. Bind the three shortcuts that matter

04. Add a custom context provider

05. Switch roles mid-edit

Next