The Easiest Way to Install MCP Servers in 2026

MCP (Model Context Protocol) lets your AI assistant call external tools -- search the web, query databases, interact with APIs. But the setup process varies wildly between clients and server types. Some require npm packages, others need Docker, and a few just take a URL. This guide covers the simplest setup path for the most popular AI clients using Scavio as an example HTTP-based MCP server.

HTTP transport is the easiest type of MCP server to configure because there is nothing to install locally. Your AI client connects to a remote URL and authenticates with a header. That is it.

Understanding MCP Transport Types

MCP servers come in three transport types:

• stdio -- the server runs as a local process. You install it via npm or pip and the client spawns it. Most complex to set up.
• SSE (Server-Sent Events) -- the server runs remotely and streams responses. Requires a persistent connection.
• HTTP -- the server runs remotely with simple request/response. Easiest to configure, works everywhere.

Scavio uses HTTP transport at https://mcp.scavio.dev/mcp, so all examples below use this approach. No npm install, no Docker, no local processes.

Claude Code

Claude Code has built-in MCP support. One terminal command:

claude mcp add --transport http scavio https://mcp.scavio.dev/mcp \
  --header "x-api-key: YOUR_SCAVIO_API_KEY"

Verify it works:

claude mcp list

That is it. Claude Code now has access to all Scavio tools -- Google search, Amazon products, YouTube transcripts, and more. You can also add it to your project's .mcp.json file so every team member gets the same tools:

{
  "mcpServers": {
    "scavio": {
      "type": "http",
      "url": "https://mcp.scavio.dev/mcp",
      "headers": {
        "x-api-key": "YOUR_SCAVIO_API_KEY"
      }
    }
  }
}

Cursor

Cursor supports MCP through its settings UI and through a config file. The fastest path is the config file. Create or edit .cursor/mcp.json in your project root:

{
  "mcpServers": {
    "scavio": {
      "type": "http",
      "url": "https://mcp.scavio.dev/mcp",
      "headers": {
        "x-api-key": "YOUR_SCAVIO_API_KEY"
      }
    }
  }
}

Alternatively, go to Settings > Tools & MCP > + Add New MCP Server, select HTTP transport, and enter the URL and header.

VS Code

VS Code's MCP support uses a .vscode/mcp.json file in your project root:

{
  "servers": {
    "scavio": {
      "type": "http",
      "url": "https://mcp.scavio.dev/mcp",
      "headers": {
        "x-api-key": "YOUR_SCAVIO_API_KEY"
      }
    }
  }
}

Open the Command Palette and run MCP: List Servers to confirm the connection. Note that VS Code uses servers as the top-level key instead of mcpServers.

Claude Desktop

Claude Desktop reads MCP config from its settings file. On macOS, edit ~/Library/Application Support/Claude/claude_desktop_config.json:

{
  "mcpServers": {
    "scavio": {
      "type": "http",
      "url": "https://mcp.scavio.dev/mcp",
      "headers": {
        "x-api-key": "YOUR_SCAVIO_API_KEY"
      }
    }
  }
}

Restart Claude Desktop after saving the file. The Scavio tools will appear in the tools menu at the bottom of the chat input.

Common Issues

• Tools not appearing: restart the client. Most read MCP config on startup only.
• Auth errors: header must be x-api-key (lowercase) with a valid key.
• Timeouts: ensure outbound HTTPS to mcp.scavio.dev is allowed.
• Wrong config key: VS Code uses servers, Claude/Cursor use mcpServers.

For the full setup guide, see the MCP integration docs.