Model Context Protocol (MCP) for AI Security

Model Context Protocol (MCP) Setup

The Model Context Protocol (MCP) allows your AI tools (like Claude Desktop) to connect directly to GridinSoft Inspector services. This enables your AI to answer the critical question: "Can I trust this site?" through real-time domain analysis and URL scanning.

Prerequisites

• A GridinSoft Inspector account and your API Key.
• An MCP-compatible client (Claude Desktop, Cursor IDE, Windsurf, etc.).
• Node.js (v18+) installed on your system.

{
  "mcpServers": {
    "gridinsoft": {
      "command": "npx",
      "args": ["-y", "@gridinsoft/mcp-inspector"],
      "env": {
        "GRIDINSOFT_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}

{
  "mcpServers": {
    "gridinsoft": {
      "command": "npx",
      "args": ["-y", "@gridinsoft/mcp-inspector"],
      "env": {
        "GRIDINSOFT_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}

Available MCP Tools

Once connected, your AI assistant will have access to the following security tools:

Recommended AI Questions

Try asking your AI these specific questions to get the most detailed security reports:

• "Perform a full security analysis of this website: example-site.com"
• "Check the trust profile and reputation for the domain gridinsoft.com"
• "Is this redirect link safe to visit? http://bit.ly/xxxx"
• "Give me a detailed report on who hosts this domain and what is its reputation."

Authentication

Authentication is handled via the GRIDINSOFT_API_KEY environment variable. This key is used to authenticate your requests and track credit usage.

Custom Implementation

If you are building your own MCP client or integrating GridinSoft services into a custom application, you can use our HTTP endpoints directly. All requests must include an Authorization: Bearer YOUR_API_KEY header.

Domain Check API Details

The domain/check endpoint provides a fast security assessment by merging real-time scans with GridinSoft's historical database.

Request Format (JSON)

{
  "apikey": "YOUR_API_KEY",
  "domain": "google.com"
}

Response Fields

Our implementation follows the official MCP HTTP Transport specification.

MCP Tool Examples

GridinSoft MCP server provides several tools that AI assistants can use to help with security tasks. below are the details and example payloads for each tool.

1. inspect_domain

Retrieves detailed information about a domain, including its security category and threat level.

Example Input:

{
  "name": "inspect_domain",
  "arguments": {
    "domain": "google.com"
  }
}

Example Output:

{
  "content": [
    {
      "type": "text",
      "text": "{
  \"domain\": \"nusewin.cc\",
  \"reputation\": 1,
  \"verdict\": \"Scam Website\",
  \"hosting\": { \"org\": \"AS13335 Cloudflare, Inc.\", \"country\": \"US\" },
  \"whois\": { \"registrar\": \"NICENIC INTERNATIONAL GROUP\", \"creationDate\": 1768427775 },
  \"site\": {
    \"title\": \"Nusewin: Crypto Casino\",
    \"tags\": { \"Scam - High Risk\": { \"risk\": true }, \"Casino\": { \"risk\": false } },
    \"ai_text_detection\": { \"verdict\": \"Human\" },
    \"ssl_certificate\": { \"protocol\": \"TLS 1.3\", \"issuer\": \"WE1\" }
  }
}"
    }
  ]
}

2. scan_url

Scans a specific URL for redirects, phishing, and malware signatures.

3. get_balance

Returns the current credit balance and subscription plan details for the authenticated user.

4. MCP Prompts

Prompts are suggested request templates that help AI understand how to interact with the user and tools.

List Prompts (/prompts):

{
  "prompts": [
    {
      "name": "website-safety-analysis",
      "description": "Deep check for phishing, scams, and threat tags",
      "arguments": [
        {
          "name": "site",
          "description": "The URL or domain to analyze",
          "required": true
        }
      ]
    },
    {
      "name": "domain-trust-report",
      "description": "Detailed profile: reputation, registrar, and hosting",
      "arguments": [
        {
          "name": "domain",
          "description": "The domain name to verify",
          "required": true
        }
      ]
    }
  ]
}

5. MCP Resources

Resources allow the AI to read reference data or documentation provided by the server.

Read Resource Content (/resources/read?uri=...):

{
  "contents": [
    {
      "uri": "gs://guide/threat-categories",
      "mimeType": "text/markdown",
      "text": "# GridinSoft Threat Categories\n\nOfficial reference for security verdicts.\n\n- **Safe** (ID: -1) - Safe\n- **Phishing** (ID: 7) - Threat/Suspicious\n- **Scam Website** (ID: 9) - Threat/Suspicious\n..."
    }
  ]
}

MCP Usage Statistics

Monitor your MCP tool execution and credit consumption below. Statistics are updated in real-time as your AI assistant interacts with the GridinSoft MCP server.

Current Limits