Rémi KokabiNovember 4, 20266 min read
Claude Code integration with lemlist through MCP, with prompts to copy
You’re in Claude Code, staring at a CSV of segmented leads or a list of accounts that just triggered a hiring signal. The data is right there, parsed and ready. Your campaigns live in lemlist, three tabs away, waiting for someone to build sequences by hand. Nobody connects the two, so the insight sits in a file while the outreach stays stuck in a queue of manual clicks.
This gap is exactly what a claude code lemlist integration through MCP closes. Once both MCP servers run in the same client, Claude Code can read your files, write scripts, and call lemlist directly, turning a terminal session into a campaign launcher.
In this article, you’ll get the setup steps and prompts you can copy right now: bulk campaign creation from a file, sizing your sending infrastructure, and auditing every active campaign for copy that’s gone stale. You’ll also get the guardrails to keep an agent from doing damage at scale. By the end, you’ll know exactly what to paste into Claude Code to get lemlist working for you instead of waiting for you.
Why connect lemlist and Claude Code through MCP
Claude Code runs in your terminal, where it can read local files, write scripts, and call any MCP server you connect. When the lemlist MCP is one of those servers, the agent can go from a CSV on your machine to a live campaign without you switching tabs. That combination turns Claude Code into the right place for bulk outbound work that would take hours in a browser.
MCP (Model Context Protocol) is an open standard that lets AI agents call external tools. An MCP server exposes a set of actions, and any compatible client can discover and call them over a single connection. lemlist’s MCP server exposes actions covering lead search, enrichment, sequence building, and campaign management. If you want a deeper look at what MCP changes for outbound teams, read the full breakdown on MCP for sales.
What you need before you start
The lemlist MCP connected in Claude, ChatGPT, or Cursor. Follow the step-by-step guide on how to set up the lemlist MCP if you haven’t done this yet.
Claude Code’s file editing and command execution tools connected in the same client, so the agent can read your local files and act in lemlist within one conversation.
Permissions limited to what the use cases below actually need. Start with read-only prompts (audits, reports) before letting the agent create or update anything in either tool.
1. Create campaigns in bulk from a file
You segmented your list into twelve verticals, and each one needs its own campaign with a tailored sequence. Building twelve campaigns by hand means repeating the same clicks dozens of times, and the copy starts drifting by campaign five.
Paste this prompt into your client with the file attached:
From this CSV of 12 segments, create one lemlist campaign per segment with its own sequence and import the leads.
The agent reads your CSV through Claude Code, parses each segment, then calls the lemlist MCP to create a campaign, build a sequence, and import the matching leads for every segment. It logs each result so you can see which campaigns were created and how many leads landed in each.
Before running this on a full list, check that your CSV columns map to the right lemlist lead fields and that no segment exceeds the size you want. You can read more about how many leads per campaign to size your segments.
2. Size and set up sending infrastructure
Planning how many domains and inboxes you need for a given daily volume is math that most reps get wrong. One miscalculation and you’re either burning a domain or underusing your infrastructure.
Paste this prompt:
I need to send 500 emails a day. Size the domains and inboxes, check availability, and set them up with warm-up in lemlist.
The agent calculates how many inboxes you need based on safe sending limits, then checks domain availability through Claude Code. After you confirm the plan, it provisions the mailboxes in lemlist and enables warm-up so each inbox builds reputation before your first campaign goes out. Keeping volume under 50 emails a day per inbox, ideally 30 to 40, protects deliverability. At 100+ emails a day per inbox, reply rates drop to 2.4% with 8.3% bounces (lemlist internal data). For a deeper look at choosing and rotating cold email domains, check the full guide.
Before letting the agent provision anything, confirm the domain names and the number of inboxes it proposes. A wrong domain purchase is hard to undo.
3. Audit every active campaign
Campaigns launched weeks ago keep running with outdated subject lines, broken links, or generic copy that never got personalized. Nobody audits them until reply rates crater.
Paste this prompt:
List all active lemlist campaigns, flag steps with links, long subjects or no personalization, and write the report to a file.
The agent pulls every active campaign through the lemlist MCP, reads each sequence step, and flags the problems: tracked links that hurt deliverability, subject lines over a safe character count, and steps missing personalization variables. It then uses Claude Code to write a structured report to a local file you can share with your team.
Before running this across all campaigns, test on a single campaign first to make sure the flags match what you actually want to fix.
Guardrails when an agent works in lemlist and Claude Code
Giving an agent write access to two tools at once is powerful, but it needs boundaries. Follow these before you let any prompt run on real data:
Review every campaign the agent creates before you launch it. The agent can build the campaign and the sequence, but a human should read the copy and confirm the audience.
Test on 10 leads first. Run the prompt on a small batch, check the result in lemlist, then scale. Campaigns of 10 to 49 leads reply at 10.6% in the US against 1.6% past 1,000 leads (lemlist internal data), so small batches are good practice anyway.
Keep unsubscribes and exclusion lists active. The agent does not override these, but verify they are on before any import.
Decide who on the team can connect the MCPs. Not every rep needs write access to both tools. Limit MCP connections to the people who will review what the agent changes.
Log what the agent changed. After each session, save the conversation or the report file. If something breaks, you need a trail.
Run it in lemlist
lemlist’s MCP server connects Claude, ChatGPT, and AI agents to 650M+ contacts, multichannel campaigns, and automated outreach workflows. Here is what the agent can do once connected:
Search the People database with 650M+ contacts and 65M+ companies, filtered by role, industry, company size, tech stack, and intent signals.
Enrich leads with waterfall enrichment across 14+ providers to find verified emails and phone numbers.
Create campaigns and sequences across email, LinkedIn, calls, and WhatsApp from a single prompt.
Read the unified inbox and campaign stats to surface what needs attention.
Act on intent signals like hiring changes, funding rounds, or website visits to trigger outreach at the right moment.
You can explore the full list of lemlist MCP use cases to see what other teams are building.
Final takeaway
Claude Code gives you file access, scripting, and terminal power. lemlist gives you the outbound engine. Connecting both through MCP means the agent can go from a local file to a launched, multichannel campaign without you copying data between tabs.
Launch AI outbound campaigns
Automate prospecting and personalize AI-powered sequences to book more meetings.