A VS Code extension that turns any OpenAI-compatible LLM endpoint into a coding assistant, configured through a graphical settings window.
Features
Graphical settings window (Gencode NG: Open Settings) for the API endpoint, API key, model, system prompt, temperature, max tokens, timeout, streaming and custom HTTP headers.
Test connection and Fetch models buttons so the configuration can be verified before use.
Chat panel (Gencode NG: Open Chat) with token streaming, cancellation and optional inclusion of the current editor selection.
Editor context menu actions: explain the selection, or ask for a refactor.
Works with OpenAI, Azure OpenAI gateways, OpenRouter, Together, Groq, Ollama (http://localhost:11434/v1), LM Studio, vLLM and anything else exposing /chat/completions.
Requirements
VS Code 1.90 or newer
Node.js 18+ to build the extension
Supported on macOS and Windows (also works on Linux); no OS-specific setup is required.
Then press F5 to launch an Extension Development Host, and run Gencode NG: Open Settings from the Command Palette.
Security notes
The API key is stored with VS Code SecretStorage (the OS keychain), never in settings.json and never in the webview state.
Endpoints must use https://; plain http:// is only accepted for localhost.
Webviews run under a strict Content-Security-Policy with a per-load nonce, and model output is inserted as text, never as HTML.
Settings
Setting
Description
llmAssistant.endpoint
Base URL of the OpenAI-compatible API
llmAssistant.model
Model identifier
llmAssistant.temperature
Sampling temperature (0–2)
llmAssistant.maxTokens
Maximum tokens per response
llmAssistant.systemPrompt
System prompt for every conversation
llmAssistant.requestTimeoutMs
Request timeout in milliseconds
llmAssistant.stream
Stream responses token by token
llmAssistant.extraHeaders
Extra HTTP headers sent with each request
The API key is not a setting; use the settings window or Gencode NG: Clear Stored API Key.
