Sort LinkedIn connection requests automatically with n8n

The short answer

Note: listing and accepting received invitations through Finchreach is coming soon. Until then, accept invitations on LinkedIn yourself; the workflow below uses tools that are live today.

n8n can handle the follow-up on your LinkedIn connection requests. You accept the invitations you want on LinkedIn; then an n8n AI Agent, with Finchreach added through the MCP Client Tool node, checks your recent notifications, looks up each new connection, compares them with rules you set (your industry, roles you care about) and drafts a short welcome message for the ones that fit. Nothing is sent yet: with approval mode on, the default, every message waits in your Finchreach approval queue until you approve it.

MCP (Model Context Protocol) is a standard plug that lets AI apps use outside tools. Finchreach is a remote MCP server for LinkedIn at https://mcp.finchreach.com/mcp. n8n’s MCP Client Tool node supports Streamable HTTP with Bearer or Header auth, so it connects directly.

Why automate LinkedIn invitations?

Pending invitations pile up. Some are people you want in your network: prospects, peers, people who met you at an event. Others are spam or sales sequences. Accepting is the quick part; the follow-up is what usually gets skipped, and a new connection with no first message goes cold fast.

A workflow does the second pass for you. Once you have accepted the good ones, it reads each new connection’s profile, writes a one-line reason for whether a welcome message makes sense, and drafts one where it does. You spend a minute approving a short list instead of writing every message yourself.

How do I build the n8n workflow?

1. Accept the invitations you want on LinkedIn, as you do today.
2. Add a Schedule Trigger node (for example every weekday at 8:00), or a Form Trigger where you paste the names or profile URLs of the people you just accepted.
3. Add an AI Agent node with a chat model of your choice.
4. Under the agent’s Tools, add an MCP Client Tool node pointing at Finchreach (setup below).
5. In the agent’s system message, write your rules for who gets a welcome message and the tone of that message.
6. Optionally, send the agent’s final output to Slack, email or wherever you read summaries, with a link to your approvals page.

During a run, the agent calls linkedin_get_notifications to see recent activity on your account, and uses linkedin_get_profile or linkedin_search (people) to check a person’s headline, current role and location. For the ones that fit your rules, it queues a welcome message with linkedin_send_dm. Every write returns pending_approval and an action id.

Finchreach doesn’t schedule LinkedIn actions. The n8n trigger decides when the workflow runs and drafts; a message is sent when you approve it.

What should the agent’s instructions say?

A system message like this gives the agent clear rules:

Ask your assistant“You follow up on my new LinkedIn connections with the Finchreach tools. I accept invitations on LinkedIn myself. Check my recent notifications and the names I give you, and look up each person’s profile. Queue a welcome DM if the person works in B2B software, logistics or supply chain, or if they mentioned a real interaction with me. Keep it under 300 characters, reference their role, no pitch and no link. Skip anyone whose headline is empty or looks like a sales sequence. End with a list: name, headline, decision, one-line reason.”

If you’d rather review decisions before anything is queued, split it into two runs:

Ask your assistant“Look up the people I accepted on LinkedIn today and tell me which ones you would message under my rules, and why. Don’t queue anything.”

Then a second, manually triggered workflow queues the messages you confirm.

How do I connect Finchreach to n8n?

The full walkthrough is in the n8n guide. The essentials:

1. Create a Finchreach account, connect your LinkedIn account on the Accounts page, and create an API key (it starts with lk_).
2. In the AI Agent node, go to Tools → + → MCP Client Tool.
3. Set the Endpoint URL to https://mcp.finchreach.com/mcp and the transport to HTTP Streamable.
4. For Authentication, choose Bearer Auth and paste your key, or Header Auth with name Authorization and value Bearer lk_....
5. Under Tools to Include, pick the tools this workflow needs, for example linkedin_get_notifications, linkedin_get_profile, linkedin_search, linkedin_send_dm and linkedin_get_action.

{ "name": "Authorization", "value": "Bearer <FINCHREACH_API_KEY>" }

Older versions of the node only support SSE, so update n8n if you don’t see the Streamable HTTP option. It works on both self-hosted n8n and n8n Cloud, as of October 2026; see the MCP Client Tool docs. Our quickstart covers the Finchreach side.

Limiting Tools to Include is worth the extra minute. An agent that can only see the tools it needs is easier to predict and can’t wander into posting or commenting.

What happens after the workflow runs?

You get a list of pending actions on the approvals page: each welcome message, with the exact text and recipient. Approve the ones you agree with and reject the rest. Approved actions run; rejected ones never touch LinkedIn. If your workflow wants to report results, the next run can check earlier action ids with linkedin_get_action.

Only message people you have already accepted on LinkedIn, so every message goes to someone who is a connection.

Automating a personal LinkedIn account carries risk under LinkedIn’s User Agreement. The approval queue lowers that risk because you review and pace every action, but it doesn’t remove it. Run the workflow at a human cadence and keep welcome messages genuinely personal.

n8n and LinkedIn: common questions

Can the workflow send connection invitations too? Yes. Finchreach has linkedin_send_invitation with an optional note up to 300 characters, also queued for approval. Outbound invitations are a different use case with more risk; start with following up on incoming ones.

Can it list or accept my received invitations? Not yet; that is coming soon. For now you accept or ignore invitations on LinkedIn, and the workflow handles the follow-up.

Does n8n see my LinkedIn password? No. LinkedIn is connected in the Finchreach dashboard; n8n only stores the API key, which you can revoke at any time.

Using another automation tool? See the Dify inbox triage workflow or the LangChain comment drafting agent. New to MCP? Start with what a LinkedIn MCP server is.

Welcome every new connection without writing each message: sign up for Finchreach and follow the n8n guide.