​
Prerequisites

​
Required Environment Variables

  • OXYLABS_USERNAME: Your Oxylabs account username.
  • OXYLABS_PASSWORD: Your Oxylabs account password.

​
Steps

  1. Install Meshagent: 
    pip install "meshagent[all]"
  2. Sign Up and Authenticate
  3. Start the MCP Oxylabs Server in a Temporary Room 
    meshagent service test \
  --room=test \
  --role=agent \
  --image=meshagent/mcp-oxylabs:latest \
  --env MESHAGENT_PORT=8001 \
  --env OXYLABS_USERNAME=YOUR_USERNAME_HERE \
  --env OXYLABS_PASSWORD=YOUR_PASSWORD_HERE \
  --port="num=8001 path=/webhook liveness=/ type=meshagent.callable" \
  --name=mcp-oxylabs-service-test
    • This command launches a temporary Room with the Oxylabs MCP server live as a callable service. When the Room closes or is inactive, the service and resources are auto-removed.
  4. Start a Chatbot with the MCP Oxylabs Toolkit in the Room 
    meshagent chatbot join \
  --room=test \
  --agent-name=mcp-oxylabs \
  --name=mcp-oxylabs \
  --toolkit=mcp-oxylabs
    • This chatbot can use all the MCP tools provided in the room. The output will include a link to join the Room.
  5. Test in the Browser
    • Visit the Room link provided in the previous step.
    • Send a message to the agent (e.g., ask it to scrape a page, Google something, or scrape Amazon) to interact live with the MCP Server tools.

​
Project Level Deployment: Persistent Services

For production and always-available tools in all project Rooms, deploy persistent MCP agent and chatbot services at the project level. These agents automatically join every new Room in the project, so users never have to manage manual tool starts.
  1. Create Persistent MCP Oxylabs Agent 
    meshagent service create \
  --role=agent \
  --image=meshagent/mcp-oxylabs:latest \
  --env MESHAGENT_PORT=8001 \
  --env OXYLABS_USERNAME=YOUR_USERNAME_HERE \
  --env OXYLABS_PASSWORD=YOUR_PASSWORD_HERE \
  --port="num=8001 path=/webhook liveness=/ type=meshagent.callable" \
  --name=mcp-oxylabs-service
  2. Create Persistent Chatbot Agent with Toolkit 
    meshagent service create \
  --image="meshagent/cli:latest" \
  --port="num=9001 path=/agent liveness=/ type=meshagent.callable participant_name=mcp-oxylabs-chatbot" \
  --env="MESHAGENT_PORT=9001" \
  --name="mcp-oxylabs-chatbot-service" \
  --command="meshagent chatbot service --agent-name=mcp-oxylabs-chatbot --toolkit=mcp-oxylabs"
Now, every new Meshagent Room in your project will have instant access to the MCP server tools—no local deployment required.

​
Tools Available

The MCP Oxylabs server provides the following tools for chatbots and agents in Meshagent Rooms:
ToolShort Description
amazon_product_scraperScrape Amazon products.
amazon_search_scraperScrape Amazon search results.
google_search_scraperScrape Google Search results.
universal_scraperGet content of any webpage.

​
Tool Details

​
amazon_product_scraper

Scrape Amazon products. Supports content parsing, user agent types, domain, geolocation, locale parameters, output formats, Amazon-specific parameters like currency, and more accurate pricing via auto-select variant. Parameters:
  • query (string): Keyword to search for.
  • autoselect_variant (boolean optional): Accurate pricing; set to true.
  • currency (string optional): Display currency.
  • domain (string optional): Domain localization (e.g., co.uk, us, fr).
  • geo_location (string optional): Target location (ISO-3166/region).
  • locale (string optional): Language header (e.g., en-US, de-AT).
  • output_format (string optional): Output: links, md, html.
  • parse (boolean optional): If false, uses output_format.
  • render (string optional): Headless browser option (html).
  • user_agent_type (string optional): Alters User-Agent header.
This tool is read-only.

​
amazon_search_scraper

Scrape Amazon search results. Supports parsing, user agent types, pagination, domain, geolocation, locale, Amazon-specific parameters, output formats. Parameters:
  • query (string): Keyword to search for.
  • category_id (string optional): Search in category.
  • currency (string optional): Display currency.
  • domain (string optional): Domain (e.g., co.uk, us, fr).
  • geo_location (string optional): Target location.
  • locale (string optional): Language header.
  • merchant_id (string optional): Specific seller.
  • output_format (string optional): links, md, html.
  • pages (integer optional): Number of pages.
  • parse (boolean optional): If false, uses output_format.
  • render (string optional): Headless browser.
  • start_page (integer optional): Starting page.
  • user_agent_type (string optional): User-Agent header.
This tool is read-only.

​
google_search_scraper

Scrape Google Search results. Content parsing, user agent types, pagination, domain, geolocation, locale, output formats. Parameters:
  • query (string): URL-encoded keyword.
  • ad_mode (boolean optional): Use Google Ads source if true.
  • domain (string optional): Localization (e.g., co.uk).
  • geo_location (string optional): Target location.
  • limit (integer optional): Results per page.
  • locale (string optional): Language header.
  • output_format (string optional): links, md, html.
  • pages (integer optional): Number of pages.
  • parse (boolean optional): If false, uses output_format.
  • render (string optional): Headless browser.
  • start_page (integer optional): Starting page.
  • user_agent_type (string optional): User-Agent header.
This tool is read-only.

​
universal_scraper

Get content of any webpage. Browser rendering, parsing, output formats. Parameters:
  • url (string): Target website URL.
  • geo_location (string optional): Target location.
  • output_format (string optional): links, md, html.
  • render (string optional): Headless browser.
  • user_agent_type (string optional): User-Agent header.
This tool is read-only.

​
Meshagent & Resources

Deploy Oxylabs MCP tools instantly in Meshagent chatrooms—cloud-native, persistent, and secured by Room or Project. Meshagent makes AI collaboration and scraping easy, scalable, and production ready!