home / mcp / begagnad mcp mcp server

Begagnad MCP MCP Server

An MCP server for Sweden's second hand marketplaces

Installation

Add the following to your MCP client configuration file.

Configuration

{
  "mcpServers": {
    "bjesus-begagnad-mcp": {
      "url": "https://begagnad-mcp.bjesus.workers.dev/sse",
      "headers": {
        "TRADERA_APP_ID": "YOUR_APP_ID",
        "TRADERA_APP_KEY": "xyz789"
      }
    }
  }
}

Begagnad MCP is an MCP server that lets AI agents search Sweden’s Blocket and Tradera marketplaces and return unified listing data such as titles, descriptions, prices, images, seller info, and direct links. It enables you to query multiple marketplaces through a single, consistent data format for use in copilots and AI assistants.

How to use

You can access the Begagnad MCP server either via a remote HTTP endpoint or by running a local MCP instance. The server provides tools to search both Blocket and Tradera, as well as combined searches, and returns a consistent data structure that includes item details, price in SEK, location, images, seller info, and a direct listing URL.

To use it from an MCP client, select the Begagnad MCP server configuration and point your client to one of the available connection methods documented in the configuration section below. When you run searches, you can request queries like "Find a specific item" or "Show me listings matching a keyword" and your client will receive unified results that you can present to end users.

How to install

Prerequisites: Node.js and npm, or a compatible runtime capable of running MCP stdio or HTTP configurations.

# Ensure you have Node.js and npm installed
node -v
npm -v

# If you are using the public remote endpoint, you don’t need to install the server locally.
# If you want to deploy your own instance, follow the deploy steps below.

Additional configuration and usage notes

Configuration is provided below for both the public remote endpoint and a local stdio setup. The public endpoint is available at the following URL and can be used directly by your MCP client. If you choose to deploy locally, run the local server as described in the deployment steps and update your client configuration to point to your deployment URL.

{
  "mcpServers": {
    "begagnad": {
      "type": "http",
      "name": "begagnad",
      "url": "https://begagnad-mcp.bjesus.workers.dev/sse",
      "args": []
    }
  },
  "envVars": []
}

Security and credentials

Tradera API credentials can be configured as Cloudflare secrets if you deploy in that environment. The following secrets are mentioned for authentication: TRADERA_APP_ID and TRADERA_APP_KEY. If you deploy locally or in another environment, store credentials securely using your platform’s secret management features.

Example usage scenarios

Example prompts you might give to Claude or another MCP client include: - Find a Linksys router with OpenWRT installed - Search for a red pickup truck under 20000 SEK - Show me vintage furniture in Stockholm

Data format

Returned data includes: item ID, title, description, price (SEK), location, images (URLs), seller name and rating, direct link to the listing, and the source marketplace (Blocket or Tradera). The structure is unified across platforms for easy consumption by your AI workflows.

Troubleshooting

If you encounter connection or data issues, verify that the server URL is reachable and that any required API credentials are configured in your deployment environment. Check that your MCP client is configured to use the correct server name (begagnad) and that the appropriate environment variables are set if you are deploying locally.

Available tools

search_blocket

Search Blocket marketplace with a query and optional limit to control results count.

get_blocket_item

Retrieve full details for a specific Blocket listing by ad_id.

search_tradera

Search Tradera marketplace with a query and optional page parameter.

get_tradera_item

Retrieve detailed data for a specific Tradera listing by item_id.

search_both

Search both Blocket and Tradera marketplaces in parallel for a single query.