Use Continue in VS Code or JetBrains with an uncensored LLM API

Continue is an open source coding assistant for VS Code and JetBrains. Models live in a config.yaml file, and its openai provider takes an apiBase so it can use any compatible server.

Works with Wild West API.

Set it up

Sign in at wildwestapi.com/signup/ with the six-digit code sent to your email, open the dashboard and create an API key. It starts with sk-ww- and is shown in full once, so copy it somewhere safe.
Open the Continue chat sidebar and click the cog next to Local Config to open config.yaml. It lives at ~/.continue/config.yaml, or %USERPROFILE%\.continue\config.yaml on Windows.
Add a models entry with provider openai, model outlaw-1 and apiBase https://wildwestapi.com/v1.
Paste your sk-ww- key as apiKey, then save. Continue reloads the file.
Pick the new model in the chat selector.

The config entry

The docs show name, provider, model, apiBase and apiKey for an OpenAI-compatible server. The reference also lists roles and capabilities.

name: Wild West
version: 0.0.1
schema: v1

models:
  - name: Outlaw 1
    provider: openai
    model: outlaw-1
    apiBase: https://wildwestapi.com/v1
    apiKey: sk-ww-your-key-here
    roles:
      - chat
      - edit
      - apply
    capabilities:
      - tool_use
    defaultCompletionOptions:
      contextLength: 500000

Do not commit this file if it holds the key, and do not put it in a repo-level config shared with others.

What the settings do

roles says which jobs the model handles: chat, edit and apply are the ones that suit a large model. Autocomplete is also a role, but it needs a fast model and a different model type, and it was not tested here.
capabilities: tool_use tells Continue it may offer tools to this model in agent mode. Remove it if the model you chose does not take tools.
defaultCompletionOptions.contextLength caps how much Continue sends. A smaller number saves money on long sessions.

One limit to know

Treat it as chat, edit and agent use. Tab autocomplete fires on nearly every pause in typing, so on a pay-as-you-go API it would add up quickly and the round trip is slower than an editor wants. Keep autocomplete on a small local model.

FAQ

Does the old config.json still work?

The Continue docs mark the JSON format as deprecated and show YAML. Use config.yaml.

Can I keep the key out of the file?

Continue has its own secrets and variable mechanisms; see its docs. This page puts the key in directly because that is the form its docs show.

Use Cline in VS Code with an uncensored LLM APICline is an autonomous coding agent for VS Code. It has a provider type called OpenAI Compatible, which is made for servers like this one. Everything is set in the settings panel, so there is no file to edit. Use Aider with an uncensored LLM APIAider is a terminal pair programmer that edits files in your git repo. It reaches OpenAI-compatible servers through LiteLLM, so the setup is two environment variables and a model name with the openai/ prefix. Use opencode with an uncensored LLM APIopencode is a terminal coding agent. It treats any OpenAI-compatible server as a custom provider that you declare in opencode.json, and it runs the request through the AI SDK's compatible package.

Use Continue in VS Code or JetBrains with an uncensored LLM API

Set it up

The config entry

What the settings do

One limit to know

FAQ

Does the old config.json still work?

Can I keep the key out of the file?

Related

Uncensored AI models on one key