LinkedIn MCP Server: How to Control LinkedIn from Claude, ChatGPT, or Cursor
Complete English guide on connecting LinkedIn to Claude Desktop, ChatGPT and Cursor via the Salesbot MCP Server. 12 tools, step-by-step setup and 5 practical examples.
LinkedIn MCP Server: How to Control LinkedIn from Claude, ChatGPT, or Cursor
Author: Petr Kubíček, Founder & CEO Sales Robots s.r.o.
Date: June 9, 2026
Category: LinkedIn Automation, MCP, AI Tools
Keywords: LinkedIn MCP server, LinkedIn automation Claude, MCP server LinkedIn, Salesbot MCP, LinkedIn API Claude, ChatGPT LinkedIn automation, Cursor LinkedIn, LinkedIn AI agent
Imagine telling Claude or ChatGPT: "Find 50 sales directors from tech companies in Prague and add them to my outreach campaign." Ten seconds later, the task is done — no clicking, no copying, no CSV exports. That's exactly what the Salesbot MCP Server makes possible.
In this guide, I'll explain what an MCP server for LinkedIn is, how it works, how to set it up in Claude Desktop or Cursor, and why this is the biggest shift in LinkedIn automation since LinkedIn itself was founded.
What Is an MCP Server and Why Does LinkedIn Need One?
Model Context Protocol (MCP) is an open standard from Anthropic (creators of Claude) that lets AI assistants communicate with external systems through standardized tools. Simply put: MCP is like a USB-C port for AI — it connects an AI brain to any application or database.
Before MCP, you could only tell an AI assistant about your LinkedIn account. With MCP, the AI assistant can directly control it.
The Salesbot MCP Server is the first implementation of this protocol for LinkedIn automation. It connects your Salesbot account to any MCP-compatible AI client — Claude Desktop, ChatGPT (with Plugins), Cursor IDE, or your own custom AI agent.
12 Salesbot MCP Server Tools
The Salesbot MCP Server gives your AI assistant 12 tools it can call autonomously or on your command:
| Tool | Description |
|---|---|
search_linkedin_people | Searches LinkedIn profiles by filters (location, title, industry, company) |
search_google_xray | Google X-Ray search for LinkedIn profiles outside the LinkedIn API |
get_contact_profile | Loads a detailed contact profile including work history |
add_contacts_to_campaign | Adds contacts to a Salesbot campaign |
send_linkedin_message | Sends a message or connection request on LinkedIn |
approve_message | Approves an AI-generated message before sending (human-in-the-loop) |
reject_message | Rejects a message and requests regeneration |
get_pending_approvals | Shows messages waiting for approval |
get_campaign_stats | Campaign statistics (sent, accepted, response rate) |
get_contacts | List of contacts with filters and pagination |
get_campaigns | Overview of all campaigns |
create_campaign | Creates a new outreach campaign |
How to Install the Salesbot MCP Server
Prerequisites
- Salesbot Pro account (14-day free trial, no credit card required)
- Claude Desktop, Cursor, or another MCP client
- API key from app.salesbot.cz → Settings → API
Step 1: Get Your API Key
Log in to app.salesbot.cz, go to Settings → API Keys, and generate a new key. The key looks like: sb_live_xxxxxxxxxxxxxxxxxxxxxxxx.
Step 2: Install in Claude Desktop
Open the Claude Desktop configuration file:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json
Add the Salesbot MCP Server:
{
"mcpServers": {
"salesbot": {
"command": "npx",
"args": ["-y", "@salesbot/mcp-server"],
"env": {
"SALESBOT_API_KEY": "sb_live_your_key_here"
}
}
}
}
Save the file and restart Claude Desktop. You'll see a hammer icon (🔨) in the bottom left corner — this means MCP tools are connected.
Step 3: Install in Cursor IDE
Open Cursor → Settings → MCP and add a new server:
{
"name": "salesbot",
"command": "npx -y @salesbot/mcp-server",
"env": {
"SALESBOT_API_KEY": "sb_live_your_key_here"
}
}
Step 4: Verify the Installation
In Claude Desktop or Cursor, type: "Using the Salesbot tool, tell me how many active campaigns I currently have."
If everything works, Claude will respond with real data from your account.
5 Practical Use Cases
Use Case 1: Prospecting from A to Z
Prompt: "Find 30 HR managers from companies with 50–200 employees in Brno. Add them to the 'HR Brno Q3' campaign and send them a connection request focused on saving time in recruitment."
Claude calls search_linkedin_people, then add_contacts_to_campaign, then send_linkedin_message — all automatically.
Use Case 2: Human-in-the-Loop Approval
Prompt: "Check all messages waiting for approval and show them to me. For each one, tell me whether I should approve it and why."
Claude calls get_pending_approvals and then recommends approval or rejection for each message. You confirm or reject with a single word.
Use Case 3: Performance Analysis
Prompt: "Compare the performance of my last three campaigns. Which has the best response rate and why do you think that is?"
Claude downloads statistics via get_campaign_stats and performs the analysis.
Use Case 4: Cursor + Code
Developers can write custom scripts in Cursor that call Salesbot tools directly from the editor:
# Cursor Agent: Automatically filter contacts by criteria
result = await mcp.call("search_linkedin_people", {
"title": "CTO",
"location": "Prague",
"company_size": "50-200",
"limit": 100
})
Use Case 5: Daily Briefing
Morning prompt in Claude Desktop: "Give me a daily summary: how many messages are waiting for approval, what is the response rate over the last 7 days, and suggest 3 actions to improve performance."
Why MCP Is Better Than Traditional LinkedIn Automation
| Traditional Tools | Salesbot MCP | |
|---|---|---|
| Control | Clicking in UI | Natural language |
| Personalization | Templates with variables | AI generates each message uniquely |
| Integration | Export/Import CSV | Direct call from AI agent |
| Approval | Automatic or manual | Human-in-the-loop as configured |
| Reporting | Dashboard | Conversational analysis |
Safety and LinkedIn Limits
The MCP Server respects LinkedIn Fair Use Policy. Salesbot uses safe limits:
- Max. 20 connection requests per day
- Progressive sending (rate limiting)
- No data scraping outside permitted ranges
- All actions are logged in the Salesbot dashboard
Conclusion
Salesbot's LinkedIn MCP Server is the first tool that truly connects the power of modern AI assistants with LinkedIn automation. Instead of clicking through dashboards, you simply describe what you want to achieve — and AI does it for you.
Try Salesbot Pro free for 14 days at salesbot.cz — no credit card, no commitment.
Frequently Asked Questions
Does the MCP Server work with the free plan?
The MCP Server is part of Salesbot Pro (880 CZK/month). The free trial (14 days) includes full access to all 12 tools.
Do I need to know how to code?
No. Claude Desktop and ChatGPT work in natural language. Cursor is for developers, but it's not a requirement.
Is the MCP Server compatible with GPT-4?
Yes. The Salesbot MCP Server implements the standard MCP protocol, which is compatible with ChatGPT and any agent supporting MCP.
How is the API key secured?
The API key is stored only in your local configuration file. It never passes through a third party.
Can I have multiple AI agents connected at the same time?
Yes. A single API key can be used in Claude Desktop, Cursor, and your own agent simultaneously.
Petr Kubíček is the founder of Sales Robots s.r.o. and Salesbot.cz. He has spent over 5 years working on B2B sales automation and AI tools for sales teams.