LM Studio lets you run large language models locally on your machine. With MCP support (available from LM Studio 0.3.6+), you can connect Geekflare tools to any local model, giving it the ability to scrape pages, search the web, take screenshots, and look up DNS.
Setup
Open the Developer settings
In LM Studio, go to Settings → Developer and enable MCP Servers.
Edit the MCP config file
LM Studio uses a JSON config file. Open it via Settings → Developer → Edit MCP Config and add: Remote URL (Recommended)
Local via npx
{
"mcpServers": {
"geekflare": {
"url": "https://mcp.geekflare.com/YOUR_API_KEY/mcp"
}
}
}
Requires Node.js installed locally.{
"mcpServers": {
"geekflare": {
"command": "npx",
"args": ["-y", "@geekflare/mcp"],
"env": {
"API_KEY": "YOUR_API_KEY"
}
}
}
}
YOUR_API_KEY with your Geekflare API key. Save the file. Restart LM Studio
Restart LM Studio to load the new MCP server. Open a chat and verify the tools appear in the tool list.
Usage Examples
In the LM Studio chat with any local model:
Scrape a webpage
Scrape https://example.com and extract the main content as markdown.
Search for "LLM fine-tuning guide 2025" and summarize the top results.
Take a screenshot of https://lmstudio.ai and tell me what you see.
Look up the A and MX records for lmstudio.ai.
Tool calling capability depends on the local model you are using. Models with
strong function-calling support (e.g. Qwen, Mistral, Llama 3.1+) will use
Geekflare tools more reliably.
