I wanted to use my local LM Studio models inside Antigravity IDE, so I connected them through MCP.
Requirements
- Antigravity IDE
- LM Studio
- A loaded model (Gemma, Qwen, Llama, etc.)
-
uvinstalled, if not then do install it with followingbrew install uv
Start the LM Studio server:
Developer → Local Server
Default endpoint:
http://127.0.0.1:1234/v1
Method 1: Using uvx (Recommended)
Add this to your Antigravity MCP config:
{
"mcpServers": {
"lmstudio-mcp": {
"type": "stdio",
"command": "uvx",
"args": [
"git+https://github.com/infinitimeless/LMStudio-MCP"
],
"env": {
"LMSTUDIO_HOST": "127.0.0.1",
"LMSTUDIO_PORT": "1234"
}
}
}
}
Important: Use git+https://, otherwise you'll get:
Direct URL references a Git repository,
but is missing the git+ prefix
Reload Antigravity.
Method 2: Clone Locally
git clone https://github.com/infinitimeless/LMStudio-MCP.git
cd LMStudio-MCP
python3 -m venv venv
source venv/bin/activate
pip install requests "mcp[cli]" openai
MCP config:
{
"mcpServers": {
"lmstudio-mcp": {
"type": "stdio",
"command": "/bin/bash",
"args": [
"-c",
"cd ~/LMStudio-MCP && source venv/bin/activate && python lmstudio_bridge.py"
]
}
}
}
Reload Antigravity.
Test
Ask:
Use lmstudio-mcp health_check
Or:
Use lmstudio-mcp to explain this file
How It Works
Antigravity
↓
MCP Bridge
↓
LM Studio
↓
Local Model
This does not replace Antigravity's main model.
It gives Antigravity access to your local model as an MCP tool, allowing you to use Gemma, Qwen, Llama, DeepSeek, or any model loaded in LM Studio.
Repository:
Top comments (0)