# How to find and connect with leads using an mcp server
> Use an mcp server to connect AI clients like Claude or Cursor to LinkedIn for programmatic lead search and automated outreach drafting.
- HTML: https://getkinlead.com/connecting-with-leads-using-mcp-server
- Markdown: https://getkinlead.com/connecting-with-leads-using-mcp-server.md

**Short answer:** You can use an mcp server to connect AI clients like Claude or Cursor to LinkedIn lead generation tools. This lets you find prospects and draft messages using natural language prompts instead of manual searches and copy-pasting.

## What is an mcp server for lead generation

Model Context Protocol (MCP) is an open standard that lets AI models connect to external data and tools. Usually, an AI is limited to its training data or what you paste into the chat. MCP lets the AI call functions on a server to get real-time data.

In lead generation, an mcp server is a bridge between an AI client (like Claude Desktop) and a lead generation API. Instead of clicking through a dashboard, the AI calls server functions to fetch profiles or send messages based on your chat instructions. This is about giving the AI a set of hands and eyes to work with.

You can stay inside your AI chat interface to do complex work. You can ask the AI to find a list of people, filter them by a specific criterion, and draft a personalized message for each one without leaving the window. This changes the process from a manual data entry job to a high-level strategic instruction job.

> The mcp server is the plumbing. It takes your high-level instruction and turns it into the specific search query and message delivery that the lead generation tool needs to execute it.

## How to set up your AI client to use the protocol

To use MCP, you need a client that supports the protocol. Claude Desktop and the Cursor code editor are the most common choices because they are designed to read configuration files and communicate with external services.

Once the server is connected, the AI client gains new tools. The AI doesn't just talk; it can now choose these tools when your prompt requires external data or an action, like searching for a LinkedIn profile or analyzing a company's recent news. The AI decides when to use the tool based on your request.

1. **Install an mcp-compatible client.** Download and install Claude Desktop or Cursor on your machine. Make sure you are running the latest version of the software, as protocol support changes often.
2. **Locate the configuration file.** Find the app settings or the JSON configuration file used by the client. This file is usually hidden in your system's application data folder, so you might need to check the app's documentation for the exact location.
3. **Add the mcp server details.** Insert the server URL and any required API keys into the config file to establish the connection. Treat the API key like a password; keep it secure.
4. **Restart the client.** Relaunch the application to activate the new tools in the chat interface. The AI won't see the new capabilities until the client is restarted.

## How to define the perfect target lead profile

The AI is only as good as your instructions. If you're too vague, you'll get leads that don't fit your business, which wastes your LinkedIn invitation limits and your time. You have to be a specific filter.

Give the AI a strict set of attributes: job titles, company size, geography, and industry. If you have a current customer who is a perfect fit, give the AI that customer's website URL. This lets the model analyze the business model and find similar companies because it can see what success looks like in the real world.

**Vague prompt**

- Find me some business owners in the UK.
- Search for people who might need marketing help.
- Find fintech leads.

**Specific prompt**

- Find founders of seed-stage fintech startups in London with 11-50 employees.
- Find VPs of Sales at healthcare SaaS companies in the US with 50-200 employees.
- Find owners of boutique creative agencies in New York that specialize in e-commerce.

## How to automate the search and drafting process

Once the AI has the tools, you combine search and drafting into one workflow. Ask the AI to find a batch of leads based on your criteria. Once the list is returned, ask the AI to analyze the profiles to find a specific hook for each person. This is where the personalization happens.

Avoid generic messages. Use the AI to find a recent post the lead wrote or a specific skill they listed on their profile. The goal is to show you did your homework, not that you used a bot. Keep connection requests under 200 characters so they are readable and don't get flagged as spam by LinkedIn.

- **you:** Find 5 CEOs of mid-sized logistics companies in Germany and draft a personalized note for each. Focus on their recent comments about supply chain efficiency.
- **draft:** I found 5 leads. Here is a draft for Marcus Schmidt: 'Hi Marcus, saw your recent post about sustainable shipping in Hamburg. I'm working on a similar project for fleet efficiency and would love to connect.'

```
Prompt: Use the lead search tool to find 10 founders of seed-stage AI startups in San Francisco. For each lead, analyze their profile and draft a connection request under 200 characters. Mention their specific focus on [Industry] and explain why their approach to [Problem] caught my eye. Format as a table with Lead Name, Profile URL, and Draft Message.
```

## What to do when leads reply to your outreach

The AI can handle the initial response, but you need a plan for when the conversation gets real. When setting up a campaign, you choose how replies are handled. You can hand the conversation to the user at the first reply, let the AI reply until there is real interest, or let the AI reply until a meeting is booked using your scheduling link.

All replies from every campaign land in one inbox, classified by intent, with interested replies first. This means you don't have to check five different campaigns to see who is actually interested in talking to you.

1. **Set the intent classification.** When building the campaign, tell the system what constitutes an 'interested reply' for your specific service. Is it a request for pricing? A request for a demo? Or just a 'tell me more'?
2. **Define the AI's goal.** If you let the AI reply, define the ultimate goal. Is it just to gather information, or is it to get the prospect to your calendar? The AI will follow the goal you set.
3. **Review and edit the AI's work.** Even if the AI handles the conversation, you can read and edit queued campaign messages on the Actions page before they send. This is your final quality check.

## How to keep the outreach process safe and organized

LinkedIn pushes back on accounts that act like bots. To manage this, you need a system that sends at a human pace. This means each LinkedIn account sends at most one action every 5 minutes, inside daily invite and message limits that you set. The system also respects the lead's time zone.

The platform also manages the list health. Stale pending invitations are withdrawn automatically to keep the pool healthy. When LinkedIn pushes back on an account, the system pauses invites on that account and emails you immediately so you can adjust your strategy.

- Always set conservative limits for your daily invitations and messages.
- Always use the send window settings to respect the lead's local business hours.
- Never try to automate more than the system allows, even if you feel you can manage it manually.

## Doing this with Kinlead

Kinlead provides an mcp server so Claude, ChatGPT, Cursor, and other clients can find leads and draft answers. You can manage your pipeline without leaving your AI environment. The server uses LinkedIn's standard search, so you don't need a Sales Navigator seat. Use your AI client to search for leads, check replies in your [unified inbox](/features/unified-inbox), and draft responses to interested prospects.

Kinlead sends from your account at a human pace. You can use the MCP server to strategize and draft while the system manages the timing and limits. The platform keeps leads in lead groups, allowing you to point any campaign at any group you've created.

## Related

- [MCP server features](https://getkinlead.com/features/agent-api-and-mcp.md)
- [AI lead search](https://getkinlead.com/tools/find-leads.md)
- [Unified inbox](https://getkinlead.com/features/unified-inbox.md)

## Common questions

**Do I need a LinkedIn API key to use MCP?**

No. You connect your AI client to an mcp server, which interacts with a lead generation tool's API to perform the actions.

**Can I use MCP with any AI chatbot?**

You can use it with any client that supports the Model Context Protocol, such as Claude Desktop and Cursor.

**Will using an MCP server get my LinkedIn account restricted?**

MCP is a communication method. Account safety depends on whether the tool you use respects LinkedIn's limits and human behavior patterns.

[Create a Kinlead account](https://getkinlead.com/signup) or [book a call](https://getkinlead.com/demo) to talk through pricing and setup.
