B2B Data MCP Server — Email, Phone & LinkedIn Enrichment

B2B Data MCP Server — Email, Phone & LinkedIn Enrichment ✅ No API Key

A single Model Context Protocol (MCP) server that gives your AI agent 13 B2B data tools: find and verify work emails, enrich emails and phone numbers, pull a LinkedIn person profile by URL or find one by name and company, search LinkedIn people, companies and company employees, read Crunchbase profiles, search the web and extract structured data from any page. Connect it to Claude, Cursor, VS Code, n8n or any MCP client with your Apify token — there is no separate account or API key to obtain.

✨ Why this Actor

• 13 tools behind one endpoint. One connection instead of wiring a dozen separate integrations, so an agent can go from a company name to a verified email and a phone number inside a single conversation.
• You pay for work done, not for requests sent. Billing happens after the lookup completes: a completed lookup is billed whether or not it finds a match, and a request that fails — invalid arguments, an upstream error, a timeout — is not billed at all.
• LinkedIn searches are billed per batch of results. Until September 29, 2026 the people and employee searches bill the search and the profiles separately. From September 30, 2026 all three LinkedIn searches bill per batch of up to 25 results returned. See the pricing section below.
• No key management. Authentication is your Apify token. Nothing else to sign up for, rotate or store.
• Read-only. Every tool is a lookup. Nothing is written, sent or modified anywhere.

🔧 How to connect

The server runs in Standby mode and speaks streamable HTTP at:

https://neuralverge--neuralverge-mcp.apify.actor/mcp

Pass your Apify API token as a bearer token in the Authorization header.

Claude Desktop / Claude Code / any client that takes a JSON config:

{
  "mcpServers": {
    "neuralverge": {
      "command": "npx",
      "args": [
        "-y", "mcp-remote",
        "https://neuralverge--neuralverge-mcp.apify.actor/mcp",
        "--header", "Authorization:${AUTH_HEADER}"
      ],
      "env": { "AUTH_HEADER": "Bearer <YOUR_APIFY_TOKEN>" }
    }
  }
}

Cursor / VS Code / clients with native remote MCP support:

{
  "mcpServers": {
    "neuralverge": {
      "url": "https://neuralverge--neuralverge-mcp.apify.actor/mcp",
      "headers": { "Authorization": "Bearer <YOUR_APIFY_TOKEN>" }
    }
  }
}

n8n: add an MCP Client node, transport HTTP Streamable, endpoint as above, and a header credential Authorization: Bearer <YOUR_APIFY_TOKEN>.

Then ask your agent something like "find the work email for the head of marketing at stripe.com and verify it" — it will pick the tools itself.

📦 What a tool returns

Every tool answers with one text block that holds a single JSON object, so an agent can read it as prose or parse it:

{
  "session_id": "8ce7ed3b-…",
  "kind": "email_finder",
  "domain": "example.com",
  "first_name": "Jane",
  "last_name": "Doe",
  "human": "# Email finder\n\n- **Found:** Yes\n- **Email:** jane@example.com …",
  "machine": {
    "found": true,
    "email": "jane@example.com",
    "confidence": "low",
    "provider": "GOOGLE",
    "message": "Verified mailbox exists"
  },
  "total_points": 10
}

• kind names the tool, and the arguments you sent are echoed next to it.
• human is a short Markdown summary; machine holds the structured fields.
• LinkedIn tools put the record count in machine.total_items and the first record in machine.first_item; the search tools also return every record in machine.items. run_crunchbase_company returns one flat machine.company object.
• run_search returns a ranked results list instead of human and machine.
• total_points is the upstream cost unit of the call. On this Actor you are billed by the Apify price table below, not by points.
• A failed call comes back with isError: true and a plain-language reason, and is not billed.

🧰 Tools and prices

Tool	What it returns	Price
run_email_validation	Deliverability verdict, mailbox existence, catch-all, provider, confidence	$0.01 / call
run_extract	Structured data from one URL, driven by plain-language instructions and an optional JSON schema	$0.02 / call
run_search	Ranked Google results (rank, title, URL, snippet)	$0.02 / call
run_email_finder	Work email for a person at a company	$0.03 / call
run_email_enrichment	Person and company profile behind an email address	$0.03 / call
run_linkedin_email	Full LinkedIn person profile from a profile URL, with a work email when one is available	$0.03 / call
run_linkedin_domain	LinkedIn profile of a person, found by full name plus company name or domain	$0.03 / call
run_linkedin_company_search	LinkedIn company search by name, industry, size, location	$0.03 / call until Sep 29, 2026; then per batch, see below
run_crunchbase_company	Crunchbase company profile: description, founders, investors, funding rounds, headcount range, industries, acquisitions	$0.04 / call
run_phone_enrichment	Owner and carrier data for an international phone number	$0.05 / call
run_phone_enrichment_us	Deeper US phone data set	$0.20 / call
run_linkedin_people_search	LinkedIn people search with title, seniority, location filters	$0.15 / search page (25 profiles) + $0.02 / profile until Sep 29, 2026; then per batch, see below
run_linkedin_company_employee	Employees of a given company, same filters	$0.04 / search + $0.02 / profile until Sep 29, 2026; then per batch, see below

🗓 New LinkedIn pricing from September 30, 2026

From September 30, 2026 the three LinkedIn search tools are billed per batch of up to 25 results returned: 1–25 results is one batch, 26–50 is two, 51–75 is three, and so on. A completed search that returns nothing is billed as one batch. There is no separate per-profile fee any more.

Tool	Per batch of 25	With scraperMode: "full"
run_linkedin_company_search	$0.15	$0.30
run_linkedin_people_search	$0.15	$0.30
run_linkedin_company_employee	$0.20	$0.20

Example: a people search that returns 60 short profiles is three batches, $0.45. One call is billed for at most 40 batches (1,000 results), and larger requests are trimmed to 1,000. All other tools keep their current prices.

Pricing

Pay per event. There is no run fee and no start fee — you pay per tool call, and only once the call has completed.

• Single-result tools are billed once per completed lookup, whether or not a match is found.
• Until September 29, 2026, run_linkedin_people_search is billed per search page of up to 25 profiles requested — asking for 100 profiles buys four pages — plus $0.02 for each profile actually returned. A search that matches no one is billed for the page only.
• Until September 29, 2026, run_linkedin_company_employee is billed $0.04 per completed search plus $0.02 per employee returned.
• Failed requests are never billed: invalid or missing arguments, upstream errors and timeouts all cost nothing.
• Connecting, listing the tools and reading tool descriptions are free.

Apify Free plan

Free-plan usage is not paid to the developer by Apify, while every call still costs money upstream, so the Free plan is a trial rather than a tier:

• 20 tool calls per user per calendar month across the whole server.
• run_linkedin_people_search: one trial call per month, up to 10 profiles.
• run_linkedin_company_employee: one trial call per month, up to 5 profiles.
• run_linkedin_company_search: one trial call per month, up to 10 companies.
• When an allowance is used up the tool answers with a plain message saying so — nothing fails and nothing is billed. The allowance resets on the 1st of each month.
• Any paid Apify plan removes all of these limits.

Prices are the same on every Apify plan.

Speed and limits

• Typical call: 3–15 seconds. The slowest tool (profile lookup by name and company) can take up to ~45 seconds.
• The server keeps a small request queue so bursts are smoothed instead of failing; a very large parallel fan-out will be served sequentially rather than dropped.
• Standby containers sleep when idle, so the first call after a quiet period is slower than the rest.

FAQ

Do I need an account or an API key besides Apify? No. Your Apify token is the only credential.

Why are the LinkedIn searches priced per batch of 25? The results come from a source that sells them in pages of 25. From September 30, 2026 you pay for the batches you actually receive, and a search that finds nothing costs one batch, because the upstream search was still run. Repeating an identical search that matched no one is answered from a short-lived cache instead of buying the page again.

What happens if a lookup finds nothing? You get an explicit "no match" answer, and the call is billed as a completed lookup — the search was still run. Only requests that fail outright are free.

Can I use the tools individually instead? Yes — the same lookups are also published as separate Actors in Apify Store if you prefer datasets over an agent connection.

Is anything written or sent on my behalf? No. Every tool is read-only.

Which MCP clients are supported? Any client that speaks streamable HTTP MCP, including Claude, Cursor, VS Code, Windsurf, n8n and custom agents built on the MCP SDKs.

🔗 Related Neuralverge Actors

Other Actors from Neuralverge that pair well with this one:

• LinkedIn People Search — search LinkedIn people by company, title and industry as a standalone Actor.
• Email Finder — a work email from a name + company domain.
• Reverse Phone Lookup — a phone number in, name, email and location out.
• AI Web Scraper — any URL to JSON by a natural-language prompt.
• More Neuralverge Actors — the full catalogue (LinkedIn, contact data, company registers, reviews): apify.com/neuralverge.

---

Disclaimer: this Actor returns data from public and licensed sources for legitimate business use. You are responsible for using the returned data in compliance with applicable law, including data protection rules in your jurisdiction.