YAML Metadata Warning: empty or missing yaml metadata in repo card (https://huggingface.co/docs/hub/model-cards#model-card-metadata)
MCP Browser Automation Server
An MCP (Model Context Protocol) server that enables AI agents to interact with websites through browser automation, with specialized support for Facebook Messenger.
Features
Messenger Tools
messenger_list_conversations- List recent conversationsmessenger_search_conversation- Search for conversationsmessenger_send_message- Send text messagesmessenger_read_messages- Read conversation messagesmessenger_send_file- Send file attachmentsmessenger_get_info- Get conversation info
Web Tools
web_navigate- Navigate to URLsweb_screenshot- Capture screenshotsweb_click- Click elementsweb_type- Type into inputsweb_extract_text- Extract page textweb_execute_js- Run JavaScriptweb_wait_for- Wait for elementsweb_get_url- Get current URLweb_accessibility_snapshot- Get accessibility treebrowser_close- Close browser
Installation
npm install
npm run build
Configuration
- Copy
.env.exampleto.env - (Optional) Set
EDGE_USER_DATA_DIRto use your existing Edge profile with saved logins
To find your Edge profile path:
- Open Edge and go to
edge://version - Copy the "Profile Path" value
Usage
With Antigravity AI
Add to your MCP server configuration:
{
"mcpServers": {
"browser": {
"command": "node",
"args": ["c:/Users/User/Desktop/VSCode/MCP-Server/dist/index.js"]
}
}
}
Testing with MCP Inspector
npm run inspector
First Time Setup
- Run the server - Microsoft Edge will open
- Log in to Messenger manually in the browser window
- The server will now be able to interact with your Messenger
Notes
- Browser runs in visible mode for debugging/monitoring
- Uses Microsoft Edge (Chromium-based) by default
- Messenger selectors may need updates if Facebook changes their UI
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support