Skip to main content
Glama
macgregordev

LeadMarina MCP server

by macgregordev

LeadMarina MCP server

Verified local-business leads, inside your AI assistant. Search any niche in any US city, get each business's emails checked for deliverability and its phones checked for line type and carrier, then send the leads to Close, GoHighLevel, Google Sheets or a CSV, Excel or JSON file. You never leave the chat.

This is a remote MCP server. There is nothing to install or run:

https://leadmarina.com/api/mcp

This repository holds the listing: this README, client config examples, the MCP registry's server.json and smithery.yaml. The server itself is LeadMarina's hosted service.

What it does

Ask in plain language, for example:

  • "Find roofing contractors in Austin, TX with a mobile number."

  • "Which plumbers I've already collected have a safe email and over 50 reviews?"

  • "Send this search to Close."

  • "Every Monday, search HVAC contractors in Dallas and Fort Worth and email me a CSV."

  • "Estimate dentists across Colorado and Utah, then run it with a 2,000-lead cap."

Each lead is one local business: its listing (name, category, address, rating, reviews, website), up to 3 emails each marked safe, risky or invalid, up to 3 phones with line type and carrier, the owner's name where one can be found, and its social profiles.

Related MCP server: Sales Prospector

Connect

You need a LeadMarina account. Every plan includes the MCP server, the free plan too: sign up.

OAuth (recommended). Add the server URL to your client. The first time a tool runs, your browser opens LeadMarina's sign-in and consent page. Nothing needs to be copied or pasted.

API key. Create a key on the API Keys page and send it as Authorization: Bearer YOUR_API_KEY. The key spends your leads, so keep it private and out of shared or committed config files.

Client set-up, with full files in examples/:

Client

How

Claude (claude.ai or the desktop app)

Customize → Connectors → + → Add custom connector → paste the server URL. OAuth runs on first use, and the connector then works in the mobile app too.

Claude Desktop (config file)

examples/claude-desktop.json: a local bridge to the remote URL, OAuth in the browser

Claude Code

claude mcp add --transport http leadmarina https://leadmarina.com/api/mcp

ChatGPT (developer mode)

Settings → Plugins → Developer mode → create an app with the server URL, authentication OAuth

Cursor

One-click install, or examples/cursor-mcp.json in .cursor/mcp.json

VS Code

examples/vscode-mcp.json in .vscode/mcp.json

Windsurf

examples/windsurf-mcp.json in ~/.codeium/windsurf/mcp_config.json

Anything else

Any MCP client that speaks streamable HTTP: the URL above, plus OAuth or the bearer header

Tools

Tool

What it does

Spends leads

search_leads

Live search for local businesses by niche and US city. Returns the first batch and keeps finding and verifying more in the background.

yes, 1 per lead

get_search_results

A search's leads with verification filled in as it progresses. Poll until enrichment is complete.

no

enrich_domains

Verify up to 10 business websites you already have.

yes, 1 per domain

query_leads

Filter and sort every lead you have collected, across all searches.

no

export_leads

Send a search, a campaign or a filtered set to Close, GoHighLevel, Google Sheets, or a CSV, Excel or JSON file.

no

list_searches

Your recent searches.

no

get_balance

Leads left, and when they reset.

no

list_views

Your saved views (named filter sets) from the leads table.

no

list_integrations

Which export destinations are connected and ready.

no

list_automations

Your scheduled searches.

no

create_automation

Re-run a search on a schedule and export each run.

yes, each run

update_automation

Pause or resume an automation.

no

delete_automation

Delete an automation. This can't be undone.

no

find_places

US states and places, with the codes an advanced search takes.

no

estimate_advanced_search

What an advanced search would find and how long it would take, without starting it.

no

run_advanced_search

Many keywords across the whole US, whole states or single places, each business charged once.

yes, 1 per unique lead

list_advanced_searches

Your advanced searches and their progress.

no

get_advanced_search

One advanced search: progress by keyword, top places, the batch running now.

no

control_advanced_search

Pause, resume or stop an advanced search.

no

Every tool carries MCP annotations: read-only tools say so, and delete_automation is marked destructive. The tools that spend leads say so in their descriptions, so your assistant can confirm with you first.

Pricing and limits

One lead is one delivered business, with all of its verification included. Lookups, reads and exports are free. See pricing and limits.

Support

License

The files in this repository (docs and config examples) are MIT-licensed; see LICENSE. Use of the LeadMarina service is governed by the Terms of Service.

Related MCP Connectors

Related MCP Servers

  • A
    license
    B
    quality
    D
    maintenance
    Enables AI assistants to search for business leads using FlashLeads' web harvest capabilities. Provides access to company contact information including emails, phones, and social profiles through natural language queries.
    3
    7 npm
    MIT
  • A
    license
    Not graded
    quality
    D
    maintenance
    Enables AI assistants to automate sales prospecting by finding contacts by role and industry, enriching data with emails and tech stacks, scoring against ideal customer profiles, and generating personalized outreach sequences. Streamlines lead generation and sales engagement workflows through integrated research and sequence generation tools.
    MIT
  • A
    license
    A
    quality
    D
    maintenance
    Enables AI tools to search and enrich B2B leads, including finding professional emails, company profiles, and filtering people and companies by various criteria.
    5
    386 npm
    MIT
  • A
    license
    Not graded
    quality
    A
    maintenance
    Enables an AI agent to scout local businesses, instantly re-rank cached leads under new scoring weights, explain each score variable by variable, run one-off website audits, and generate ready-to-send outreach pitches or WhatsApp openers from natural language. It exposes the same discovery/scoring engine as the CLI, so results can be listed, scored and opened in a dashboard without leaving the chat.
    1
    MIT