Use the LeadsPlease® MCP Server Try It Use Cases Install Free
The mailing list MCP server · Live now

Mailing List MCP Server. Plug 250M+ consumer records into Claude, Cursor & ChatGPT.

The first MCP server for mailing lists, direct mail, and B2B prospecting. Lets Claude, Cursor, ChatGPT, Windsurf, and any MCP-compatible AI agent count, preview, order, and fulfil targeted U.S. mailing lists in plain English — from 250M+ consumer households and 18M+ verified businesses. One copy-paste config; thirty-second install; nothing to build.

Install now — it's free → See a live demo
Live in 30 seconds No credit card on free tier USPS CASS-certified data SOC 2 · GDPR · CCPA
250M+
Consumer records
18M+
Business contacts
650+
Demographic selects
5
MCP tools shipped
30s
Install time

💡 Building a UI for non-developer users? The DataWidget® for Web-to-Print is the same data, surfaced as an embeddable three-panel UI (chat + map + filters) instead of an MCP-tooled agent. Pair with the MCP Server when you want both — agent prefills the criteria, widget lets the user fine-tune visually.

Five tools your AI agent gets

Every MCP tool exposes a clear, typed schema with natural-language descriptions, so agents call them correctly the first time. No documentation hunt; no broken JSON.

count_audience

How many records match?

Pass any combination of geography (city, county, ZIP, polygon, radius), demographics (income, age, homeowner, length of residence, vehicle, pets, …), or B2B firmographics (SIC, NAICS, employees, revenue). Returns the exact record count plus tier-pricing breakdown.

FREE TIER · the hook
preview_audience

Show me 10 sample rows

Returns 10 redacted sample records with name initials + city + selected demographics, so the agent (and you) can sanity-check audience composition before spending a cent. PII never leaves our servers in this call.

FREE TIER · trust builder
order_list

Buy the list and charge me

Places the actual order on the customer's saved card. Returns an order ID, line-item price, and ETA. Optional approval webhook so a human confirms above-threshold orders before they bill.

Metered · revenue moment
fulfil_list

Where's my CSV?

Polls the order status; when ready, returns a signed download URL or pushes the file straight to S3, Salesforce, HubSpot, or any HTTPS endpoint the caller passes in. Async-aware so agents don't deadlock.

Metered · delivery
describe_filters

What's the catalog of filters I can use?

Returns the full hierarchy of demographic, firmographic, and behavioural selects available, with their accepted values and example queries. Self-documenting — lets agents say "You can also filter by length-of-residence, propensity-to-move, accredited-investor status, vehicle make…". Eliminates hallucinated filter names.

FREE TIER · discovery

Works with every major MCP host

Drop the same config block into whichever AI tool your team uses. Same five tools, same auth, same data. We track the MCP spec releases and ship updates within a week.

🤖Claude Desktop ⚡Claude Code 🎯Cursor 🌊Windsurf 💬ChatGPT (desktop) 🔗n8n 📦Smithery.ai ⚙Any MCP-compatible agent

Why an MCP, not just a REST API

LeadsPlease® has had a REST API for years — teams build entire workflows on it. MCP is the next layer: the same data, but consumable by AI agents directly. The difference is in distribution and time-to-first-call.

VectorREST APIMCP Server
Discovery Customers find you, read docs, write integration code. Listed in Claude / Cursor / Smithery marketplaces. Install in 30 seconds.
First call Days of integration. SDK choice, auth flow, response parsing. One copy-paste config block in an AI host's settings file.
Who calls it Your customer's developers, then their app. Your customer's AI agent (Claude, GPT, Cursor) calls it directly — no glue code.
Discoverability of features Read the docs, hope the field names are right. describe_filters tool means agents auto-discover what's possible at runtime.
Stickiness App-level integration. Switching costs are moderate. Workflow-level integration. Once embedded in an AI stack, switching cost is high.
Best for Predictable, high-volume, structured workflows. AI-mediated workflows where the agent decides what to query and when.

Use both. Most of our customers run the REST API for their predictable bulk-load jobs and the MCP server for everything that an AI agent or sales person triggers in natural language.

See the MCP in motion

On the left: a real Claude Desktop conversation, captured live, calling our MCP tools. On the right: a working count_audience form so you can hit our endpoint right now — same call your AI agent will make.

Claude Desktop · with LeadsPlease® MCP Recorded
Sales Ops
Find me HVAC contractors in the Phoenix metro — 5 to 50 employees, doing $1M–$10M revenue, and family-owned. I need a count and an idea of cost.
Claude
I'll use the LeadsPlease® MCP to count that audience.
▶ count_audience
database: "business"
location: "Phoenix, AZ" · radius_mi: 25
sic_codes: ["1711", "1731"] // HVAC + electrical contractors
employees: { "min": 5, "max": 50 }
revenue_usd: { "min": 1000000, "max": 10000000 }
flags: ["family_owned"]
I found 847 matches. Pricing comes back at $0.42/record for this volume tier — ~$356 total. Want me to pull a 10-record preview before you commit?
Sales Ops
Yes, preview them.
Claude
▶ preview_audience
audience_id: "aud_8f23a7" · limit: 10
Sample rows:
1. D**** AC Services · Mesa, AZ · 12 employees · ~$2.4M rev · family-owned ✓
2. S******* Heating & Cooling · Phoenix, AZ · 27 employees · ~$5.1M rev · family-owned ✓
3. A*** HVAC · Tempe, AZ · 8 employees · ~$1.3M rev · family-owned ✓
… 7 more

Audience looks clean — all in metro, all in revenue band, all family-owned. Ready to order?

Sales Ops
Yes — order it and push it to our HubSpot.
Claude
▶ order_list
audience_id: "aud_8f23a7"
format: "csv"
deliver_to: "hubspot://default"
▶ fulfil_list
order_id: "ord_9b14cc" · poll: true

Done. Order ord_9b14cc charged $356.74 to your default card, and 847 contacts just landed in your HubSpot under list "HVAC Phoenix Family-Owned · 2026-05-02". Total round-trip: 11 seconds.

Run a real query right now

No signup. Same count_audience tool your AI agent will call — counts only, no PII returned. Hit it from this page or hit https://mcp.leadsplease.com/mcp/v1/count directly.

—
matching records
The MCP call your agent would make
// MCP tool call from any host (Claude, Cursor, ChatGPT, …)
tool: "count_audience",
arguments: {
  database:  "consumer",
  location:  "Phoenix, AZ",
  radius_mi: 10,
  filters: {
    income_min_usd: 100000,
    homeowner:      true
  }
}

What people actually do with the MCP

Six recurring patterns we see across early-access customers. Each replaces hours of manual list-building or weeks of integration work with a thirty-second prompt.

🎯
Sales-ops automation
RevOps · SDR teams

The pattern: a sales lead asks Claude for a target list inside a meeting. The list is in their CRM before the meeting ends.

Pull me 1,000 commercial roofers in the Tampa metro doing $2M–$10M revenue, and load them into Salesforce as a campaign called "Roofing Q3 push".
Outcome
Count + preview + order + push-to-Salesforce in one Claude turn. Replaces a 4-step ZoomInfo + manual-export workflow.
Lift
30× faster first-list time vs. manual research.
📨
Marketing automation pipeline
Marketing-ops · n8n / Zapier teams

The pattern: every new content asset triggers an n8n workflow that asks the LeadsPlease® MCP to build the matching audience and ship it to the email or direct-mail platform.

When a new blog post is tagged "homeowner-finance", build a 50k-record list of high-income homeowners in our top 8 metros and queue an email send.
Outcome
Audience generation moves from a quarterly batch job to a per-asset trigger. Lists always match the content.
Lift
5× pipeline contribution vs. evergreen lists.
🏢
ABM look-alike enrichment
Enterprise sales · ABM

The pattern: point an agent at your closed-won list. It infers ICP, asks the MCP for similar businesses (industry, size, geo, revenue band), and hands you a ranked target list.

Look at our last 50 closed-won deals. Build a 500-account look-alike list with the same SIC codes, employee bands, and U.S. regions, prioritised by revenue.
Outcome
ICP discovery + audience build + scoring in one prompt. No data-science engagement required.
Lift
2× meeting-set rate vs. cold lists in early customer pilots.
🏠
Real-estate farming agent
Realtor / mortgage marketing

The pattern: an agent every Monday pulls fresh "likely-to-list" homeowners (length-of-residence 7+, equity-tap signals, recent valuation searches) within each of a realtor's farm zips, and queues a postcard drop.

For each of these 5 zips, pull homeowners with LOR ≥ 7 years, drop the file into the Way Better Postcard print queue, and email me the CSV summary.
Outcome
Recurring direct-mail program runs on autopilot. Realtor reviews + approves; no manual list-pulling.
Lift
Goes from 1 mailer/quarter to 1 mailer/week per realtor.
🔬
Market-research agent
Strategy · Product · Founders

The pattern: ask Claude what the addressable market for a persona looks like, by geo. The MCP returns counts across slices in seconds — no panel survey, no analyst.

What does the U.S. addressable market for empty-nester homeowners earning $150k+ look like, broken down by top-25 metros? Show counts and the size of each demographic sub-slice.
Outcome
TAM / SAM / SOM analysis in one prompt. A research deck's first slide built in 90 seconds.
Lift
What used to take a strategy consultant a week.
🌐
Franchise list-building automation
Franchise networks · Multi-location

The pattern: corporate publishes one playbook agent that spawns territory-scoped lists for every franchisee on a schedule — brand-mandated filters baked in, local geography per location.

Every Monday, for each of our 142 locations, pull a fresh new-mover list within their 5-mile trade area and email it to the franchisee.
Outcome
One agent + one MCP replaces 142 manual list-pulls. Brand consistency guaranteed.
Lift
Eliminates a 0.5-FTE ops role at corporate.

Common questions

If your question isn't answered here, the LeadsPlease® team typically replies inside the same business day.

What is the LeadsPlease® MCP Server?

An official MCP (Model Context Protocol) server that lets AI agents in Claude Desktop, Claude Code, Cursor, ChatGPT desktop, Windsurf, n8n, and any other MCP-compatible host call five tools — count_audience, preview_audience, describe_filters, order_list, fulfil_list — to build, price, and order targeted U.S. mailing lists from 250M+ consumer and 18M+ business records.

How do I install the LeadsPlease® MCP Server in Claude Desktop?

Edit ~/Library/Application Support/Claude/claude_desktop_config.json on macOS (or %APPDATA%\Claude\claude_desktop_config.json on Windows), paste the snippet from the Install tab, and restart Claude Desktop. Total install time: 30 seconds.

Is there a free tier?

Yes. The Free tier covers evaluation and prototypes — 100 count_audience calls per month, 10 preview_audience calls per month, unlimited describe_filters. Use the public demo key lp_demo_2026 to try the protocol from your AI host of choice without signing up.

How is this different from the LeadsPlease® REST API?

The REST API is for application integrations: predictable, high-volume, structured workflows. The MCP server is for AI-mediated workflows where the agent itself decides what to query. Same data; different audience. Most customers use both — the API for batch jobs, the MCP for natural-language conversations and agent workflows.

Which AI tools can use the MCP Server?

Any MCP-compatible host. Today that includes Claude Desktop, Claude Code, Cursor, Windsurf, ChatGPT desktop (Plus / Team / Enterprise), n8n via the community MCP node, Smithery.ai, plus any custom client built on the Model Context Protocol SDK from Anthropic.

How do I get an API key?

Use the public demo key lp_demo_2026 for free-tier evaluation, or sign up on the Install tab to receive a personal lp_live_… key by email in under a minute. Builder ($99/mo) and Scale (custom) tiers unlock unlimited count and preview calls plus order fulfilment.

What kind of mailing lists can the MCP build?

Consumer mailing lists (250M+ U.S. adults), business mailing lists (18M+ U.S. businesses by SIC/NAICS), new-mover lists, new-homeowner lists, turning-65 / T65 lists, absentee-owner lists, veterans lists, saturation/occupant lists — with 650+ demographic and firmographic selects. Call describe_filters to discover the full catalogue at runtime.

Does it cost anything to call count_audience?

No. Counts and previews are free under the free tier (100 + 10 calls/month). You only pay when you actually order a list via order_list, which charges your saved card at the per-record tier returned by count_audience. Free-tier callers cannot place orders; that's gated to paid tiers.

How does white-label work for resellers?

Resellers in the LeadsPlease® program can publish a branded MCP at {reseller}.mcp.leadsplease.com on the Scale tier. Their AI-agent customers connect to that subdomain; orders flow through the reseller's Stripe Connect account; the 80/20 split fires automatically — same plumbing as the white-label DataWidget®.

Is the data USPS-verified and CCPA-compliant?

Yes. All consumer mailing lists are USPS CASS-certified and refreshed every 7–30 days. The full LeadsPlease® database meets CAN-SPAM, CCPA / CPRA, GDPR, and TCPA requirements. Records returned via the MCP carry the same compliance and accuracy guarantees as orders placed through the website or the REST API.

Install in 30 seconds

Pick your AI tool. Copy the config block. Paste it into the host's settings. Restart. Done. The free tier covers most evaluation work; upgrade when you start ordering lists.

Step 1 of 2

Drop this into your AI tool

~/Library/Application Support/Claude/claude_desktop_config.json Claude Desktop · macOS / Windows / Linux 
{
  "mcpServers": {
    "leadsplease": {
      "command": "npx",
      "args": ["-y", "@leadsplease/mcp-server"],
      "env": { "LEADSPLEASE_API_KEY": "lp_live_…" }
    }
  }
}
Pricing

Pay only when you start ordering

Free

$0
forever

For evaluation, prototypes, internal demos, and personal accounts.

  • 100 count_audience calls / month
  • 10 preview_audience calls / month
  • Unlimited describe_filters
  • Standard support (email)
  • No order_list / fulfil_list
Get free key
Recommended

Builder

$99/mo

For founders, sales-ops, RevOps teams shipping AI workflows. Most popular.

  • 10,000 count_audience calls / month
  • 1,000 preview_audience calls
  • Unlimited order_list · pay list cost only
  • List records billed at our retail rate
  • Priority support · 1-business-day SLA
  • 5 API keys for team members
Start Builder

Scale

Custom

For franchises, white-label resellers, and high-volume agentic workflows.

  • Unlimited tool calls + records
  • White-label MCP under your domain
  • Per-reseller revenue share (Stripe Connect)
  • Dedicated environment / SSO / SOC 2 packet
  • Volume pricing on records
  • Enterprise SLA · named CSM
Talk to sales
Step 2 of 2

Get your API key