The Model Context Protocol is an open standard, introduced by Anthropic in November 2024, for connecting AI assistants to external tools and data. An MCP server describes its tools; the assistant decides when to call them during a conversation.

Which assistants can use the LeadMind MCP server?

Claude (as a custom connector), ChatGPT (in developer mode), Gemini CLI, and any other client that supports remote MCP servers with OAuth. The setup guide has the steps for each.

Can the assistant overspend my credits?

A job only ever holds credits up to your balance, or the budget you give the assistant, and each API key has a daily credit cap you set. Credits are charged only for delivered leads.

Does LeadMind see my conversation?

No. The assistant sends only the arguments of each tool call, such as the prospecting prompt or a job id. LeadMind never receives your chat history or the assistant’s memory.

MCP prospecting: lead generation inside Claude, ChatGPT and Gemini

MCP prospecting means asking your AI assistant for leads and letting it do the legwork: start a live lead job, wait for it, read the scored results back into the chat, and carry on with the research or writing you actually wanted. This guide covers how it works and workflows that hold up in Claude, ChatGPT and Gemini.

Guide · By LeadMind AI (how we source and check data) · Updated 25 September 2026

Start free Watch the demo

How it works

You add LeadMind’s MCP server to your assistant once and sign in with OAuth (or an API key).
You ask in plain language: “find 40 gyms in Dubai Marina with phone numbers”.
The assistant calls create_lead_job. The job parses the request, picks a source and holds credits.
It polls get_job_status. If the job needs a detail (usually the location), the assistant asks you and answers with clarify_job.
On delivery it reads leads with get_leads (50 per page) or hands you a CSV link from export_leads_csv.

The same server includes the email verifier: verify_emails for up to 25 addresses inline, and start_email_verification with get_email_verification for lists of up to 5,000 (paid plans). The setup guide lists every tool.

Spending stays under your control

Credits are held, not charged, until leads are delivered; filtered rows, duplicates and failed jobs cost nothing.
A job never holds more than your balance, or the budget you name (“spend at most 50 credits”).
Each API key has its own daily credit cap, so a looping agent cannot drain an account.
OAuth access tokens expire after an hour; revoke the connector or the key at any time.

Workflow: research a market in Claude

“Find 30 interior design firms in Al Quoz with WhatsApp. Show the top 10 by score as a table with website and reason, then summarise what the best-scoring firms have in common.”

Claude runs the job, reads the leads, and does the analysis in the same thread. Add it under Customize → Connectors → Add custom connector with the URL from the setup guide.

Workflow: clean list for a campaign in ChatGPT

“Get 50 physiotherapy clinics in Toronto with email, verify the emails, and give me a CSV of only the deliverable ones.”

ChatGPT chains create_lead_job, verify_emails or a bulk verification, and export_leads_csv. Enable it under Settings → Apps & Connectors → Developer mode → Create.

Workflow: prospecting from the terminal with Gemini CLI

“Run 25 dental clinics in Jumeirah without online booking and write a one-line opener for each based on its score reason.”

Add the server URL to ~/.gemini/settings.json and run /mcp auth leadmind. The assistant drafts the openers; sending them is up to you, because LeadMind does not send messages.

Tips

Put the signals in the request (“with WhatsApp”, “no online booking”); they become filters, not just instructions to the assistant.
Name a budget for large jobs, and ask for the CSV rather than all rows in the chat.
Keep the prompt library open for phrasing that works.
Prefer code over chat? The same flow is in the REST API.

Frequently asked questions

What is MCP?: The Model Context Protocol is an open standard, introduced by Anthropic in November 2024, for connecting AI assistants to external tools and data. An MCP server describes its tools; the assistant decides when to call them during a conversation.
Which assistants can use the LeadMind MCP server?: Claude (as a custom connector), ChatGPT (in developer mode), Gemini CLI, and any other client that supports remote MCP servers with OAuth. The setup guide has the steps for each.
Can the assistant overspend my credits?: A job only ever holds credits up to your balance, or the budget you give the assistant, and each API key has a daily credit cap you set. Credits are charged only for delivered leads.
Does LeadMind see my conversation?: No. The assistant sends only the arguments of each tool call, such as the prospecting prompt or a job id. LeadMind never receives your chat history or the assistant’s memory.