Enrichment as a Tool Call: How to Connect LeadModule's MCP Server to Claude and n8n

Your AI sales agent can write personalized outreach, research company news, and draft follow-ups. What it can't do — without live enrichment — is verify whether the contact data it's working with is real.

LeadModule is a waterfall enrichment platform that cascades through 8+ data providers to find verified emails and phone numbers, stopping on the first confirmed match. It achieves ~95% find rates versus the ~60% you get from single-provider tools like Apollo or Hunter.

LeadModule's MCP server exposes that enrichment engine as tools any MCP-compatible AI agent can call natively — no HTTP boilerplate, no polling loops, no webhook wiring.

TL;DR

• LeadModule exposes three MCP tools: enrich-lead, find-leads, and create-config
• Connect to Claude Desktop in under 5 minutes using a single config block
• n8n 1.40+ has a native MCP node — wire it into any automation workflow
• BYOK keys are respected: enrichment routes through your own provider accounts at wholesale cost
• MCP endpoint: https://app.leadmodule.ai/api/v1/ai

Why AI Agents Need Live Enrichment

Most AI sales agents work from static data sources: a CSV you uploaded, a CRM sync from last Tuesday, or whatever the model was trained on. When a contact's email bounces or a LinkedIn URL goes stale, the agent has no way to recover.

The gap shows up in three places:

1. Personalization without verification. The agent drafts an opener referencing a contact who left the company six months ago.
2. Hallucinated contact data. Models fill in missing fields with plausible-looking emails that don't exist.
3. Blocked outreach. Sequences built on unverified emails hit bounce rates above 5%, triggering deliverability penalties.

Live enrichment at the tool-call level fixes this. Instead of baking contact data into the prompt, the agent calls enrich-lead when it needs a verified email — and gets one from a live provider cascade, not a cached guess.

What LeadModule's MCP Server Exposes

The MCP server is available at https://app.leadmodule.ai/api/v1/ai and registers three tools:

enrich-lead

Find a verified email or phone number for a contact.

At least one of firstName, lastName, companyDomain, or linkedInUrl is required.

find-leads

Search for new leads by title, location, company attributes, and tech stack.

create-config

Create a new waterfall configuration with a specific provider order and validators.

Connect to Claude Desktop (5 Minutes)

You need a LeadModule API key. Find it in your workspace settings under API Keys. Keys start with lm_live_.

Open your Claude Desktop config file:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json

Add the LeadModule MCP server:

{
  "mcpServers": {
    "leadmodule": {
      "type": "http",
      "url": "https://app.leadmodule.ai/api/v1/ai",
      "headers": {
        "Authorization": "Bearer lm_live_your_api_key"
      }
    }
  }
}

Restart Claude Desktop. You'll see a hammer icon (🔨) in the conversation toolbar — that indicates active MCP tools. Click it to confirm enrich-lead, find-leads, and create-config are listed.

Test it

Ask Claude:

"Find the email for Sarah Chen, Head of Growth at Notion."

Claude will call enrich-lead with the data it has, run it through your configured provider cascade, and return a verified email — or tell you none was found.

Wire LeadModule into n8n via MCP

Recent versions of n8n include a native MCP Client Tool node. This lets you use LeadModule enrichment inside any n8n AI agent workflow without writing HTTP Request nodes.

1. Add an AI Agent node

Create a workflow with an AI Agent node. Set the model to your preferred provider (OpenAI GPT-4o, Anthropic Claude, etc.).

2. Add the MCP Client Tool node

Add an MCP Client Tool node and connect it to the AI Agent node's Tools input.

Configure the node:

• Connection Type: HTTP (SSE)
• URL: https://app.leadmodule.ai/api/v1/ai/sse
• Authentication: Header Auth
• Header Name: Authorization
• Header Value: Bearer lm_live_your_api_key

3. Define the agent task

In the AI Agent node's system prompt, describe what the agent should do with enrichment:

You are a lead enrichment assistant. When given a contact's name and company,
call enrich-lead to find their verified email address. Return the email,
the provider that found it, and the confidence score.
If enrichment fails, say so — do not guess.

4. Pass leads from a trigger

Connect any input node (Google Sheets, Webhook, HubSpot trigger) to the AI Agent. Map the contact fields:

Input: { "firstName": "Alex", "lastName": "Kim", "companyDomain": "stripe.com" }

The agent will call enrich-lead, get the verified email, and pass it downstream to your CRM write or email sequence node.

End-to-End Agent Example

Here's what a complete agentic enrichment workflow looks like with MCP:

Prompt given to the agent:

"Enrich this lead: Alex Kim, Engineering Manager at Stripe. Find their email and draft a one-sentence personalized opener referencing their tech stack."

What the agent does:

1. Calls enrich-lead with { firstName: "Alex", lastName: "Kim", companyDomain: "stripe.com" }
2. Receives: { email: "alex.kim@stripe.com", confidence: 0.97, provider: "prospeo" }
3. Drafts opener: "Hey Alex — saw Stripe is expanding the Payments Infra team; curious how you're thinking about enrichment at the scale you're operating."

The enrichment happens inside the agent turn — no separate pipeline step, no context-switching.

BYOK Cost Control

If you've configured Bring Your Own Keys (BYOK) providers, MCP enrichment routes through those keys. You pay the provider directly at their published rates — LeadModule doesn't add markup.

At typical volumes:

• Prospeo: ~$0.01–0.02 per found contact
• Findymail: ~$0.01–0.03 per found contact
• Dropcontact (async): ~$0.01–0.05 per found contact

For agentic use cases where you're enriching on-demand (not bulk), BYOK keeps costs predictable. If you'd rather not manage API keys, LeadModule managed credits work identically through MCP — credits are deducted per successful enrichment.

Set forceRefresh: false (the default) to take advantage of LeadModule's enrichment cache. If the same contact was enriched recently, the cached result is returned instantly at no provider cost.

Rate Limits and Production Considerations

• API rate limit: 60 requests per minute per workspace
• Async providers: Dropcontact, BetterContact, FullEnrich, and Wiza use webhook callbacks. MCP calls to these providers block until the result returns (typically 10–60 seconds). For interactive agent use, prefer sync providers (Prospeo, Findymail, Hunter, LeadMagic) first in your waterfall config.
• Bulk workloads: MCP is designed for on-demand enrichment. For processing lists of 500+ contacts, use the REST API /api/v1/enrich endpoint with async batching — it's faster and doesn't block agent execution.
• Config selection: You can create dedicated waterfall configs for different use cases (e.g., US contacts vs. EU contacts) and pass the configId to enrich-lead to route requests accordingly.

Related Guides

• How to Build a Lead Enrichment Workflow in n8n with LeadModule
• LeadModule API Quickstart
• Waterfall Enrichment: How Provider Order Affects Find Rates