Private beta — 0/10 founder seats (-50% for life) Join the waitlist
Leadmaps

Leadmaps for Claude — European business data in your conversations

Claude speaks MCP natively. Add one config block and it can count, search and export European local businesses — counting is always free, so Claude sizes every request before spending a credit.

Setup

  1. Create your API key at app.leadmaps.io/keys (a free lm_test_ key works, no card needed).
  2. Claude Code: run `claude mcp add --transport http leadmaps https://mcp.leadmaps.io/mcp --header "Authorization: Bearer lm_live_xxx"`.
  3. Claude Desktop: add the JSON block below via mcp-remote in claude_desktop_config.json.
  4. Ask Claude to `count_places` for your target — it is free — then search or export.
{
  "mcpServers": {
    "leadmaps": {
      "type": "http",
      "url": "https://mcp.leadmaps.io/mcp",
      "headers": { "Authorization": "Bearer lm_live_xxx" }
    }
  }
}

3 copy-paste prompts

MCP tools

ToolCost
count_placesFree — always call first
search_places1 credit per place returned (30-day dedup free)
get_place1 credit (dedup)
list_categoriesFree
create_export1 credit per delivered row
get_export_statusFree

Why Claude + Leadmaps works so well

Claude is the most capable MCP client on the market: it plans multi-step research on its own. Given the Leadmaps server, a single prompt like "Build me a list of 150 real-estate agencies in Lille rated 4.4+, with a website, exported as CSV" becomes a safe, budgeted pipeline:

  1. count_places (free) — Claude checks the result size first: 612 matches.
  2. create_export — Claude launches an async export limited to 150 rows.
  3. get_export_status — Claude polls once, then hands you the download URL.
  4. 150 credits charged, and re-running the same request within 30 days is free (dedup).

Working with the spending cap

If your account hits its spending cap mid-export, the job pauses (paused_cap) instead of failing. Claude is told explicitly: "do NOT poll further; tell the user to raise the cap at app.leadmaps.io/billing then call resume". No infinite loops, no surprise invoices.

Claude Code one-liner

claude mcp add --transport http leadmaps https://mcp.leadmaps.io/mcp \
  --header "Authorization: Bearer lm_live_xxx"

Claude Desktop (via mcp-remote)

{
  "mcpServers": {
    "leadmaps": {
      "command": "npx",
      "args": ["mcp-remote", "https://mcp.leadmaps.io/mcp",
               "--header", "Authorization: Bearer lm_live_xxx"]
    }
  }
}

A prospecting agent prompt that works

You are my local prospecting agent. Always call count_places before any search. Never deliver more than 200 places without asking me. Prefer create_export for anything above 50 results. Report credits_charged after every paid call.

FAQ

Does the Leadmaps MCP work with claude.ai, Claude Desktop and Claude Code?

Yes, all three. claude.ai: custom connector pointed at https://mcp.leadmaps.io/mcp with the Authorization header. Claude Desktop: an mcp-remote block in claude_desktop_config.json. Claude Code: one-line `claude mcp add --transport http leadmaps …`.

Can Claude spend my credits without asking?

Counting, facets and export status are free; only search_places, get_place and create_export charge credits. Add “always call count_places first and ask me before delivering more than 200 places” to your system prompt — and your account spending cap remains the ultimate guard.

What if Claude runs the same export twice?

Nothing bad: any listing already delivered and billed to your account can be re-delivered for free for 30 days, across all channels. A retry or re-run of the same prompt never costs extra credits.

Can I test with Claude without a credit card?

Yes: create a free lm_test_ key at app.leadmaps.io/keys — it gives access to a sandbox of ~5,000 real listings through the MCP, for 0 credits and no card.