# A LinkedIn MCP server that does the outreach

> HeyLead is an open-source (MIT) MCP server for LinkedIn outreach. It finds the right people, writes in your voice, follows up and handles replies, from Claude, Claude Code, Cursor or any MCP client.

This is the Markdown twin of https://heylead.dev/linkedin-mcp-server (the page, in text). Index for agents: https://heylead.dev/llms.txt

Last updated 23 September 2026

HeyLead is an MCP server for LinkedIn outreach. Add it to Claude, Claude Code or Cursor, say who you want to reach, and it finds the people, writes to them in the voice of your own LinkedIn posts, follows up and handles the replies.

Most LinkedIn MCP servers give your AI raw access: read a profile, search people, send one message. HeyLead is built one level up, around the campaign. Your AI describes the audience, and HeyLead runs the outreach from start to finish.

## Connect HeyLead

Paste this into Claude Code or your AI chat:

`Set up HeyLead using https://heylead.dev/llms.txt`

Your AI will follow the setup for your client. The hosted connection needs no Python, uv or API key.

### Claude Code terminal

`claude mcp add --transport http --scope user heylead https://heylead.dev/mcp`

Open `/mcp`, select HeyLead, and follow the browser sign-in. If HeyLead is not listed, start a new session and open `/mcp` again.

### Claude Code desktop, without the CLI

If `claude` is not on PATH, paste the setup prompt above into the Code tab. Your agent can add the remote server to your Claude Code configuration without installing a CLI. It must preserve your existing settings. Start a new session, then sign in through `/mcp`.

### Claude Desktop chat, Cowork and claude.ai

Open Customize → Connectors (or Settings → Connectors), choose **Add custom connector**, and enter:

`https://heylead.dev/mcp`

Click Add, then Connect and complete browser sign-in. Enable HeyLead in your conversation. If Add custom connector is unavailable, check your Claude plan or ask your workspace administrator.

### Cursor

Add this to `.cursor/mcp.json`, merging it with existing servers, then connect and sign in:

`{"mcpServers":{"heylead":{"url":"https://heylead.dev/mcp"}}}`

### VS Code

Merge this into `.vscode/mcp.json` in the project (the key is `servers`, not `mcpServers`), then start the server from the MCP panel and sign in:

`{"servers":{"heylead":{"type":"http","url":"https://heylead.dev/mcp"}}}`

### Finish and verify

1. Sign in with Google when your client opens HeyLead in the browser.
2. Connect LinkedIn in the [dashboard](https://heylead.dev/dashboard/).
3. Ask your AI: “Show my HeyLead status.” It should call `show_status` and report your workspace, even if you have no campaigns yet.
4. Describe your product and audience, and ask for a draft campaign. Review it before asking HeyLead to launch.

A saved server configuration is only the first step. Setup is verified when HeyLead answers a tool call. If sign-in fails or tools do not appear, follow the [troubleshooting steps](https://heylead.dev/llms.txt).

## What it does

- **Finds the right people.** Turns a plain description ("CTOs at Series A fintech in London") into buyer personas and LinkedIn searches, then scores each person for fit.
- **Writes like you.** Reads your LinkedIn profile and posts and writes connection requests and messages in that voice.
- **Follows up.** Sends follow-ups after a connection is accepted and stops when someone replies.
- **Handles replies.** Reads each reply, answers questions and moves interested people toward a meeting. You can turn automatic replies off and answer every reply yourself.
- **Reports back.** Pipeline, replies and outcomes in your chat or on the web dashboard.

It works for sales, recruiting, partner and investor outreach, job search and LinkedIn posts. Selling is the most developed of these.

## The tools

HeyLead exposes tools for reading status and managing outreach. Your client discovers the available tools when it connects. The ones you will use most:

- `setup_profile`: connect your account and learn your writing voice
- `generate_icp`: build buyer personas from a description
- `create_campaign`: find prospects and save a draft campaign
- `campaign`: launch, pause, resume or emergency-stop
- `check_replies` and `inbox`: read and answer replies
- `show_status` and `analytics`: pipeline and results

The full list is in [llms-full.txt](https://heylead.dev/llms-full.txt).

## How it reaches LinkedIn

Through [Unipile](https://www.unipile.com), an independent connectivity provider. HeyLead is not a Chrome extension and does not drive a browser to use LinkedIn. On the hosted plan HeyLead provides the connection, and outreach runs from the cloud with your laptop closed. Self-hosted installs bring their own Unipile account and LLM key; see [the open-source page](https://heylead.dev/open-source-ai-sdr).

## What keeps it in check

HeyLead sends from your own LinkedIn account at a human pace: at most 20 invitations a day and 100 a week on a free LinkedIn account (more on Premium or Sales Navigator), Monday to Friday 08:00 to 22:00 in your time zone, minutes apart. It backs off when LinkedIn pushes back and resumes on its own. You can pause any campaign at any time.

- A campaign is a draft until you launch it, so you see the prospects first.
- Invitations and messages stay within daily and weekly limits for each account, with randomised timing.
- Sending happens on weekdays, 08:00 to 22:00 in your own time zone.
- `campaign(action="emergency_stop")` pauses everything at once.

LinkedIn's own rules are in the [Terms](https://heylead.dev/terms).

## Source and listings

- GitHub (MIT): [github.com/D4umak/linkedin-outreach-mcp](https://github.com/D4umak/linkedin-outreach-mcp)
- PyPI: [pypi.org/project/heylead](https://pypi.org/project/heylead/)
- Official MCP Registry: `io.github.D4umak/heylead`
- Glama: [glama.ai/mcp/servers/D4umak/linkedin-outreach-mcp](https://glama.ai/mcp/servers/D4umak/linkedin-outreach-mcp)

Related: [LinkedIn outreach from Claude](https://heylead.dev/claude-linkedin-outreach) · [Open-source AI SDR](https://heylead.dev/open-source-ai-sdr) · [What is HeyLead?](https://heylead.dev/what-is-heylead)
