Clear your LinkedIn invitation backlog with Grok Bot

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.

If you have dozens of pending LinkedIn invitations, accept the ones you want on LinkedIn and let Grok Bot do the follow-up: look up each new connection, sort them against rules you give it, and draft a short welcome message for the ones worth one. With Finchreach, each message waits in your approval queue, so you confirm everything before it is sent.

Finchreach is a remote MCP server (MCP is a standard plug that lets AI apps use outside tools) at https://mcp.finchreach.com/mcp. One honest caveat up front: as of October 2026 we haven’t found official documentation for adding custom MCP servers to Grok Bot. Third-party guides describe a chat-driven setup that accepts an API key header. If that doesn’t work for you, the Grok guide covers the documented xAI API route.

Why does an invitation backlog matter?

Pending invitations pile up quietly. Some are people you met last week and would like to keep in touch with; some are recruiters, prospects or peers who could matter later; some are clearly spam. Leaving them all unanswered means the good ones go cold, and even after you accept, a connection with no first message is easy to forget.

Following up is a judgment task with clear rules, which is exactly where an assistant helps. Finchreach gives it the live tools for this: linkedin_get_notifications shows recent activity on your account, linkedin_get_profile and linkedin_search look people up, and linkedin_send_dm queues a message. Sending is a write, so it goes to your approval queue.

What should you ask Grok Bot to do?

First, accept the invitations you want on LinkedIn. Then describe who deserves a welcome message. The clearer the rules, the better the sort:

Ask your assistant“I just accepted these people on LinkedIn: [names or profile URLs]. Also check my recent notifications. Look up each person and sort them into three groups: welcome (B2B SaaS sales, marketing or customer success, or anyone I met in person), skip (unrelated headline or obvious selling), and look closer (everything else). Show me a table with name, headline and your reason.”

Review the table, move anyone you disagree with, then let it act:

Ask your assistant“Draft a two-sentence welcome DM for everyone in the welcome group except the two recruiters. Queue them one at a time and give me the action ids.”

Each message appears at /approvals in the Finchreach dashboard. Approve them there and they run on your account at your pace. You can also ask the assistant to check progress, which uses linkedin_get_action to read each action’s status.

Ask your assistant“For the look-closer group, search LinkedIn for each person’s company and tell me in one line what the company does.”

How do you connect Grok Bot to Finchreach?

1. Sign up at finchreach.com/sign-up, connect your LinkedIn account under Accounts, and create an API key under API keys.
2. In Grok Bot, ask it to add a custom MCP server named finchreach at https://mcp.finchreach.com/mcp, and give it the header Authorization with the value Bearer followed by your key, when it asks for authentication.
3. Confirm the setup, then attach the server in a conversation with @ as the third-party guides describe.
4. Ask it to list your LinkedIn accounts to check the connection.
5. If Grok Bot won’t accept a header, use the xAI API remote MCP tool instead (see the Grok guide), or a header-capable client from the docs.

These steps come from a third-party guide (Composio), not from xAI, and we rate them low confidence as of October 2026. Your team or admin may need to approve new servers, and the server must be publicly reachable, which Finchreach is. Treat your API key like a password and revoke it in the dashboard if you share a chat by mistake.

What rules make a good follow-up filter?

• Role and industry: name the job titles and sectors you want to welcome, not just “relevant people”.
• Context: someone who wrote you a specific note usually deserves a more specific welcome.
• Mutual context: ask the assistant to search for the person’s company if the headline is vague.
• Hard no’s: list who never gets a message, such as people whose note was a cold sales pitch.
• A cap: ask for no more than a set number of messages per session, so your activity stays at a normal human pace.

You can save your rules as a short message you paste at the start of each session, so every run uses the same standard.

How personal should the welcome message be?

Very. Finchreach’s linkedin_send_dm messages a connection, and each message waits for approval. A short, specific note to someone you actually met is welcome. A templated pitch sent to everyone you accepted is the kind of thing that annoys people and draws attention to your account. If you want help with replies, see answering LinkedIn messages from your editor.

What are the risks?

Automating a personal LinkedIn account carries risk under LinkedIn’s User Agreement. Finchreach’s approval queue and human pace lower that risk, but don’t remove it.

A message can’t be unsent once it goes out, so read each draft before you approve. Accepting or removing connections is something you do on LinkedIn yourself.

Frequently asked questions

Can Grok Bot list or accept my received invitations? Not yet; listing and accepting received invitations through Finchreach is coming soon. For now, accept or ignore them on LinkedIn.

Can it send invitations too? Yes, with linkedin_send_invitation and an optional note of up to 300 characters, also approved one by one. See researching prospects before you connect.

Is Grok Bot the same as Grok on grok.com? They are separate entry points. For grok.com connectors and the xAI API, read the Grok integration page.

Where do I start? Create an account and follow the quickstart.