Purpose

Prospecting is the job nobody wants to do. Sitting on LinkedIn for two hours, opening profile after profile, reading their posts, hunting for something specific to say. The LinkedIn Lead Engine hands that busywork to Claude. It searches for people who match your ideal client, reads what they have been posting, scores how well they fit, and drafts a comment and a DM for each one, anchored to something real.

This is the list builder, not an auto-sender. Claude puts everything in a Google Sheet, and you decide who to contact and when.

Before you start

Day 1 and Day 2 done: your voice and tone profile, your icp.md, and the Project you built on Day 2
A Claude plan that supports connectors and Projects
A free Apify account (Apify gives new accounts monthly free credit)
A Google account for Google Drive and Google Sheets

How the pieces fit

The agent has four parts. Each one maps to something you either already have or set up in this guide.

Part	What it is	Where it comes from
Memory	icp.md holds who you are looking for, who to skip, and what a good lead looks like. A Google Sheet called Lead Engine is the agent's second memory, so it never shows you the same person twice.	icp.md from Day 2. The sheet is created by Claude on the first run.
Skill	linkedin-instructions.md is the method: how to search, filter, score, and write a comment and DM that does not sound like a salesperson.	Step 1 below. Use it exactly as it is.
Access	Apify to search LinkedIn and read posts. Google Drive to create and update the sheet.	Step 0 below.
Decision rules	Search with your ICP filters, read recent posts, score each lead, draft outreach, and wait for your approval.	Built into linkedin-instructions.md.

KEY IDEA

Everything that makes this yours lives in icp.md. Everything about how the job gets done lives in linkedin-instructions.md. Never edit the second file. Customize the first.

Connect Apify and Google Drive

Where this happensClaude Settings, in claude.ai or the Claude Desktop app.

You need two connectors. If either one is already connected, skip it.

Apify

Go to Settings, then Connectors, then browse connectors and search for Apify. Connect the native Apify connector.
If it is not in the directory, click Add custom connector and use the server URL https://mcp.apify.com.
Sign into your Apify account and authorize.
Enable the connector for the Project you will use.

Google Drive

In Connectors, find Google Drive and click the plus icon.
Sign in with Google and check off the permissions it asks for. It takes a few seconds.

NOTE

Need a walkthrough for Apify? The Day 3 resources include a video on connecting Apify to Claude.

Get linkedin-instructions.md

Where this happensYour browser, then your computer.

This is the skill file you see on screen in the Day 3 video. Open the doc, then save it as a .md file.

Open linkedin-instructions.md

Open the Google Doc above.
Go to File, then Download, and choose Markdown (.md) if it is offered. Otherwise copy all the text and paste it into a plain text file.
Name the file linkedin-instructions.md.

DO NOT EDIT

Use this file exactly as it is. It is the same for everyone. All your customization comes from icp.md.

Check icp.md in your Project

Where this happensThe Project you built on Day 2.

You do not need a new Project for this agent. Open your Day 2 Project and confirm that icp.md is in its knowledge files. Each agent runs as a chat inside that same Project.

Give your icp.md a quick read before the first run. The skill uses it to decide which search filters to apply, who to skip, and how to score each lead.

Run the pipeline

Where this happensA new chat started inside your Day 2 Project.

Open a new chat inside your Project.
Attach linkedin-instructions.md to the message.
Paste the prompt below and set your batch size.
When Claude asks for permission to use Apify, choose Always allow. It needs this to search for actors.

Prompt Read icp.md and the attached linkedin-instructions.md before doing anything else. Run the LinkedIn Lead Engine using those two files. Check my Lead Engine Google Sheet first, so you never return anyone who is already in it. If it doesn't exist yet, create it. For each lead: read their recent posts and profile, score the fit against my ICP, and draft one outreach message anchored to something real from their posts. Add every lead to the Lead Engine sheet with the profile link, score, and draft message. Never send anything. Batch size: [X] leads.

NOTE

Swap [X] for your batch size. Start with 3 to 5 leads for your first run while you check the scoring and the drafts. Apify charges credits per profile, and a small batch also finishes faster.

Review and reach out

Where this happensThe Lead Engine Google Sheet in your Google Drive.

When the run finishes, Claude gives you a link to the Lead Engine sheet. Open it and for each lead:

Check the fit score so you know which leads are worth your time.
Read the draft message. It should refer to something specific from their recent posts.
Edit it if you want, then click the profile link in the sheet and send it yourself.

WARNING

Claude does not send any messages. You are always the one choosing who gets contacted and when.

Running future batches

Steps 0 to 2 are one-time setup. After that, you have two ways to run new batches.

Option A: run it manually. Open a new chat inside your Project, attach linkedin-instructions.md, paste the Step 3 prompt, and set your batch size. Do this whenever you want a fresh batch.

Option B: run it on a schedule. Do Option A once and check the output. When you like the leads and the messages, stay in that same chat and ask Claude to turn it into a scheduled task. For example:

Prompt Turn this into a scheduled task that runs every Monday at 8am with a batch size of 7 leads. When it finishes, send me the link to the Lead Engine Google Sheet.

Claude builds the schedule from the run you just tested, so it already knows the files, connectors, and steps.

Frequently Asked Questions

Do I need to create a new Project for this agent?

No. Use the Project you built on Day 2. It already holds your icp.md, and each agent runs as a chat inside it.

Do I need to edit linkedin-instructions.md?

No. Use it exactly as it is. Everything that makes the agent yours comes from icp.md.

Do I have to redo the setup every time?

No. Steps 0 to 2 are one-time. Every future run is a new chat inside your Project with the file attached and the Step 3 prompt pasted in, or a scheduled task if you set one up.

How does it avoid showing me the same lead twice?

Claude creates a Google Sheet called Lead Engine and adds every lead to it. On each new run it checks the sheet first, so people already in it are skipped.

What batch size should I run?

Start with 3 to 5 leads, since Apify charges credits per profile. Once you trust the scoring and the drafts, move to 5 to 10 a week.

Does Claude send the messages for me?

No. Claude drafts the messages and puts them in the sheet. You read them, edit if needed, and send the ones you like yourself.

I'm stuck. Where do I get help?

Send a message in Slack. Our team is there to help you.

Go deeper

Want AI employees running this across your whole business?

This is one workflow. In the live cohort we build AI employees across marketing, sales, and operations, so you can run systems like this across your entire business.

Join the live cohort