OpenAI Compatible VSCode LLM Server

A VSCode extension that provides an OpenAI-compatible API server interface directly within VSCode.

Features

• Implements OpenAI-compatible REST API endpoints
• Optional auto-start configuration
• Server status monitoring via status bar
• Integrated logging system
• Compatible with OpenAI API clients

Endpoints

The server runs on http://localhost:3775 by default and provides the following endpoints:

GET /v1/models

Lists available models (currently returns mock data)

Example:

curl http://localhost:3775/v1/models

POST /v1/chat/completions

Chat completion endpoint compatible with OpenAI's API format

Example:

curl -X POST http://localhost:3775/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4o",
    "messages": [
      {
        "role": "user",
        "content": "Hello, how are you?"
      }
    ],
    "stream": true
  }'

Using the Extension

1. Install the extension in VSCode
2. Use the command palette (Ctrl+Shift+P) to access the following commands:
   • "OpenAI Server: Start OpenAI Server"
   • "OpenAI Server: Stop OpenAI Server"
   • "OpenAI Server: Show OpenAI Server Status"
   • "OpenAI Compatible Server: View Server Logs"
3. Monitor server status in the VSCode status bar
4. The server will be available at the configured port (default: 3775)

Requirements

• Visual Studio Code 1.96.0 or higher

Extension Settings

This extension contributes the following settings:

• openaiCompatibleServer.port: Port number for the OpenAI compatible server (default: 3775)
• openaiCompatibleServer.defaultModel: Default model to use when none is specified in the request. One of:
  • claude-3.5-sonnet
  • gpt-4o
  • gpt-4o-mini
  • o3-mini
  • o1 (default: gpt-4o)
• openaiCompatibleServer.autoStart: Automatically start the server when VSCode launches (default: false)

To modify these settings:

1. Open VSCode Settings (File > Preferences > Settings)
2. Search for "OpenAI Compatible Server"
3. Adjust the settings as needed

Known Issues

• The server currently returns mock responses

Release Notes

0.0.5

• Added logging system with viewable logs
• Added status bar integration
• Added auto-start configuration option
• Server port changed to 3775 by default