How to Connect Claude Desktop and Cursor to a Live B2B Lead Enrichment MCP API

#mcp #ai #automation

How to Connect Claude Desktop and Cursor to a Live B2B Lead Enrichment MCP API

Integrating real-time B2B firmographics into LLMs is now seamless via the Model Context Protocol (MCP), allowing AI agents to query live company data directly from their native interfaces. By connecting to a dedicated B2B lead enrichment MCP server, developers can provide Claude Desktop, Cursor, or VS Code with high-fidelity tools to perform deep-dive company research, technographic mapping, and intent signal tracking without writing custom API glue code.

Core Features

The B2B Lead Enrichment MCP server exposes a sophisticated toolset designed for high-accuracy data retrieval. Unlike static databases, this implementation leverages live web-crawling and data-aggregation layers to provide:

Deep Firmographics: Access real-time revenue estimates, employee counts, industry classifications, and headquarters geolocation.
Technographic Stack Mapping: Identify the specific CRM, CMS, and marketing automation tools a company currently utilizes.
Intent Signals & Market Presence: Detect recent funding rounds, hiring trends, and general market sentiment.
Hallucination-Resistant Schema: Every tool utilizes strict Zod annotations for input parameters, ensuring the LLM strictly adheres to valid JSON-RPC 2.0 structures and minimizes parameter hallucinations.

Configuration for Claude Desktop and Cursor

To give your AI agent access to these tools, add the following configuration to your claude_desktop_config.json (located at ~/Library/Application Support/Claude/claude_desktop_config.json on macOS) or your Cursor MCP settings:

{
  "mcpServers": {
    "b2b-enrichment": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-http",
        "--url",
        "https://lead-enrichment-mcp.agent-infra.workers.dev/mcp"
      ],
      "env": {
        "API_KEY": "YOUR_ENRICHMENT_API_KEY"
      }
    }
  }
}

JSON-RPC Response Example

When you trigger the enrich_lead tool through an agent like Cursor or Claude, the MCP server returns a structured payload that the LLM uses to ground its reasoning. Below is a sample response for a technographic and firmographic query:

{
  "tool": "enrich_lead",
  "result": {
    "companyName": "ExampleCorp",
    "domain": "examplecorp.com",
    "industry": "Enterprise SaaS",
    "firmographics": {
      "employees": "500-1000",
      "revenueRange": "$50M-$100M",
      "hq_location": "San Francisco, CA"
    },
    "technographics": [
      "Salesforce",
      "HubSpot",
      "Segment",
      "AWS"
    ],
    "intentSignals": {
      "hiring_velocity": "High",
      "recent_funding": "Series C"
    },
    "confidenceScore": 0.94
  }
}

Risk-Free Metered Billing for AI Agents

The biggest challenge with autonomous SDR swarms and AI research agents is the cost of low-quality data. This MCP server utilizes a Risk-Free Metered Billing model. Instead of paying for every request, your account is only debited for successful enrichments that return a Confidence Score > 0.6. If the server cannot find high-fidelity data for a specific domain, the query cost is $0, allowing you to scale your AI workflows without the risk of paying for "data not found" errors or hallucinated results.