---
title: "LinkedIn MCP servers compared: what each one can do"
description: "LinkedIn has no official MCP server. Eight third-party options compared by what they can do, how they sign in, what they cost, and how recently they changed."
canonical: "https://innovate-blog.com/articles/linkedin-mcp-servers-compared"
last-updated: "2026-11-06"
---

# LinkedIn MCP servers compared: what each one can do

> LinkedIn has no official MCP server. Eight third-party options compared by what they can do, how they sign in, what they cost, and how recently they changed.

By Moez Zhioua. Published 2026-11-06. Updated 2026-11-06. Category: AI connectors. Estimated reading time: 11 minutes.

## In brief

- There is no official LinkedIn MCP server. Checks of the MCP Registry, LinkedIn's GitHub organization and its developer documentation on 6 November 2026 found none, so every option is third-party.
- Connectors built on LinkedIn's official API (Composio, Pipedream, Zapier) can post and read your own profile, but cannot search members, send messages or send invitations.
- Servers that search, message or invite work through your own LinkedIn session, a scraper, or a browser extension. LinkedIn's User Agreement restricts that kind of software, and the account risk is yours.

## The short answer

A LinkedIn MCP server lets an AI assistant such as Claude or ChatGPT call LinkedIn-related tools through the Model Context Protocol. LinkedIn does not publish one. The eight servers compared below come from connector platforms, open-source developers, a scraping marketplace and hosted products. They differ less in what they promise than in how they reach LinkedIn.

Choose by the job. For posting, use a connector built on LinkedIn's official API. For reading profiles and search results inside your assistant, the open-source servers that drive your own browser session do the most, and carry the most account risk. For outreach campaigns run from Claude or ChatGPT, a hosted workspace keeps lists, sequences and limits outside the chat.

This page compares the servers. For setup steps by route, read the companion guide on how to connect Claude and ChatGPT to LinkedIn (/articles/connect-claude-chatgpt-to-linkedin).

## LinkedIn MCP servers at a glance

Every fact in this table was checked on 6 November 2026 against the project's GitHub repository, documentation or pricing page. Not stated means the source did not say. Maintained is the date of the latest commit in the public repository; a hosted service can change without a public commit.

*LinkedIn MCP servers compared, checked 6 November 2026*

| Server | Type | What it can do | Auth method | Cost | Maintained |
| --- | --- | --- | --- | --- | --- |
| Composio LinkedIn toolkit | Official API connector | Post, comment and delete posts; read your own profile and organization page statistics; limited lookup of other members. No member search, messages or invitations. | LinkedIn OAuth through Composio's app or your own | Free Hobby tier with 100,000 tool calls a month; Pro from $29 a month | Toolkit version 20261101; ComposioHQ/composio last commit 5 Nov 2026 |
| Pipedream LinkedIn | Official API connector | Post as a member or organization, comment, like, read member profiles and organization roles. No member search, messages or invitations. | LinkedIn OAuth through Pipedream Connect | Hosted MCP free for personal use; Connect pricing for products | LinkedIn components last commit 5 Nov 2026 |
| Zapier MCP | Official API connector | Two LinkedIn actions: Create Share Update and Create Company Update. No profile reading, search, messages or invitations. | LinkedIn OAuth inside Zapier; Zapier sign-in or token for the MCP server | Two tasks from your Zapier plan per successful tool call | Plugin repo last commit 22 Oct 2026; hosted service not public |
| stickerdaniel/linkedin-mcp-server | Self-hosted open source | Read profiles, companies, jobs, feed and inbox; search people, companies, jobs and posts; send messages and connection requests. No posting tool listed. | Your own LinkedIn session, kept in a local browser profile | Free (Apache-2.0); you run it | 4,119 stars; last commit 4 Nov 2026 |
| eliasbiondo/linkedin-mcp-server | Self-hosted open source | Read profiles, companies, company posts and jobs; search people and jobs. No messaging, invitation or posting tools listed. | Your own LinkedIn session from a login window, kept locally | Free (MIT); you run it | 212 stars; last commit 24 Oct 2026 |
| Apify LinkedIn MCP | Scraper | Extract profile data, work history and company details, plus email addresses and phone numbers where the Actor finds them. No posting, messages or invitations. | Apify OAuth or API token; no LinkedIn login on your side | Pay per result: the configured Actor charges $10 per 1,000 results | apify/apify-mcp-server last commit 5 Nov 2026; Actor update date not stated |
| Unipile MCP | Hosted SaaS, unofficial API for developers | Through the Unipile API: search people in Classic, Sales Navigator and Recruiter; read profiles; send messages, InMail and invitations; publish posts. | LinkedIn account linked by credentials, session cookie or hosted auth; MCP uses a scoped API key | From €49 ($55) a month for up to 10 accounts; 7-day trial | unipile/unipile-mcp last commit 29 Sep 2026 |
| OutreachGenie MCP (https://www.outreachgenie.ai/linkedin-mcp) | Hosted SaaS outreach workspace | Import profile URLs into lists, keep notes, build flows with visits, invitations, messages and waits, prepare and activate campaigns, draft posts. No live LinkedIn browsing or search; LinkedIn steps run through its Chrome extension. | OAuth with scopes in Claude and ChatGPT; workspace API key in Claude Code and Codex | MCP on every plan. Free: 80 outreach actions and 500 new prospects a week. Pro $15 a month; Agency $20 per LinkedIn account a month | Docs repo last commit 28 Oct 2026; server code not public |

## Is there an official LinkedIn MCP server?

No. On 6 November 2026 the official MCP Registry listed 65 server names containing linkedin, and none sat under a LinkedIn or Microsoft namespace. LinkedIn's GitHub organization had no MCP repository, and LinkedIn's developer documentation on Microsoft Learn describes OAuth APIs, not an MCP server. Treat any product that calls itself the official LinkedIn MCP with suspicion until LinkedIn says so.

The closest thing to official access is LinkedIn's own API. Its open permissions (https://learn.microsoft.com/en-us/linkedin/shared/authentication/getting-access) let any developer app sign members in and post, comment or like on their behalf. Marketing, Sales Navigator and Talent APIs need partner approval, and the compliance APIs are closed. There is no open API for searching members or sending messages, and that one fact explains most of the table above.

## Which LinkedIn MCP server fits which job

### Posting through the official API: Composio, Pipedream or Zapier

If the assistant only needs to publish or comment as you, or as a company page you administer, an official-API connector is the conservative choice. You authorize LinkedIn through OAuth, the connector stores the token, and LinkedIn's API enforces what that token can do.

Composio's LinkedIn toolkit (https://docs.composio.dev/toolkits/linkedin) exposes the most LinkedIn tools of the three, including organization page statistics. Its documentation warns that the shared OAuth app hits rate limits, so use your own LinkedIn app for production. Pipedream's LinkedIn components (https://github.com/PipedreamHQ/pipedream/tree/master/components/linkedin) are open source and were updated the day before this check. Zapier (https://zapier.com/apps/linkedin/integrations) is the narrowest, with two posting actions, but it may already be connected in your account. None of them can message, invite or search members, because LinkedIn's open API cannot.

LinkedIn's API Terms of Use also tell developers not to use the APIs to automate posting. A person who reviews and approves each post the assistant drafts is a better pattern than a schedule the assistant runs alone.

### Open source and self-hosted: stickerdaniel/linkedin-mcp-server

stickerdaniel/linkedin-mcp-server (https://github.com/stickerdaniel/linkedin-mcp-server) is the most widely used open-source option, with 4,119 GitHub stars and a commit on 4 November 2026. It runs on your computer, opens a Chromium browser signed in to your own LinkedIn account, and gives the assistant tools to read profiles, companies, jobs and the inbox, search people and posts, send messages and send connection requests. Its README says messages are sent after confirmation.

The project's own FAQ states the trade-off: LinkedIn's User Agreement prohibits automated access, accounts using automated tools can be restricted or banned, and there is no guarantee for the account. The server also depends on LinkedIn's page structure, which is why the README recommends automatic updates. eliasbiondo/linkedin-mcp-server (https://github.com/eliasbiondo/linkedin-mcp-server) is a smaller, read-only alternative built on the same browser-session approach.

### Data extraction: Apify

Apify's LinkedIn MCP page (https://apify.com/mcp/linkedin-mcp-server) connects an assistant to a community scraping Actor through Apify's hosted server. You sign in to Apify, not to LinkedIn, and pay per result. The Actor's listing says it tries to find email addresses and, for paying users, mobile numbers.

That convenience moves the responsibility to you. LinkedIn's User Agreement prohibits scraping, and personal data about other people brings data-protection duties wherever you work. The guide to LinkedIn outreach without doxxing (/articles/linkedin-outreach-without-doxxing) explains why contact details someone did not choose to share with you should stay out of a workflow, and out of a model's context.

### Building a product: Unipile

Unipile's MCP server (https://github.com/unipile/unipile-mcp) is aimed at developers whose own app will act on users' LinkedIn accounts. It describes the server as a build-time tool: the coding agent reads the API contract and tests calls against connected test accounts. Accounts are linked with credentials, a session cookie or Unipile's hosted authentication rather than LinkedIn's OAuth, so it carries the same platform risk as other session-based tools, priced per connected account.

### Outreach campaigns from Claude or ChatGPT: OutreachGenie

OutreachGenie's hosted server gives the assistant a prospecting workspace rather than LinkedIn itself. Claude or ChatGPT can create lists, import profile URLs, write notes, build flows of visits, invitations, messages and waits, and pair a list with a flow as a campaign draft. Starting a campaign needs a separate OAuth scope that you can leave out.

The drawbacks are real. LinkedIn actions run later through OutreachGenie's Chrome extension in your own signed-in browser, which has to stay open while a campaign runs. The server does not browse or search LinkedIn live or return general LinkedIn data, so it is a poor fit for research. The Free plan caps outreach at 80 actions and new prospects at 500 a week. It is also a newer product: its public MCP repository was created in October 2026. LinkedIn's rules on automation software apply to its extension as much as to any other tool.

Its tool list, scopes and plan limits are documented on the OutreachGenie LinkedIn MCP page (https://www.outreachgenie.ai/linkedin-mcp).

## What LinkedIn's rules say

Section 8.2 of LinkedIn's User Agreement (https://www.linkedin.com/legal/user-agreement), effective 3 November 2025, lists things members agree not to do. They include using software, scripts, robots, crawlers or browser plugins to scrape or copy the Services; using bots or other unauthorized automated methods to access the Services, add or download contacts, or send or redirect messages; and using someone else's account, including by copying cookies.

LinkedIn's help page on prohibited software and extensions (https://www.linkedin.com/help/linkedin/answer/a1341387) says third-party tools that scrape, modify or automate activity on the site are not permitted and that accounts using them can be restricted or shut down, and its page on automated activity (https://www.linkedin.com/help/linkedin/answer/a1341543) says a restricted member is asked to disable the tool before the account is re-enabled. A server that signs in as you and works through LinkedIn's pages is acting with your account, so the restriction lands on your profile, not on the server's developer. No vendor can promise otherwise, and slower pacing does not turn prohibited automation into permitted use.

Official-API connectors avoid that particular problem because LinkedIn issued the token, but they can only do what the open API allows. Every option still needs the controls in the access checklist for AI agents (/articles/what-ai-agent-needs-before-it-can-act): narrow scopes, approval before anything is sent, a record of what ran, and a way to stop.

## How this comparison was checked

On 6 November 2026 I read each project's README or documentation, its pricing page where one exists, and GitHub metadata through GitHub's API: stars, licence and the date of the latest commit. I did not install every server, run them against a LinkedIn account or measure restriction rates, so nothing here is a performance test or a rating.

Two kinds of candidates were left out. adhikasp/mcp-linkedin, a server often listed in older roundups, returned a 404 on GitHub. The many smaller servers in the MCP Registry were not reviewed one by one. Several of them wrap LinkedIn's official API for posting; one, Dishant27/linkedin-mcp-server (https://github.com/Dishant27/linkedin-mcp-server), notes in its README that LinkedIn documents no public endpoint for general people search.

If the assistant will help with outreach, hold it to the standard you would use without AI: the review-before-send workflow (/articles/ethical-linkedin-outreach-best-practices) and the consent and opt-out guidelines (/articles/linkedin-outreach-consent-guidelines). For tools without an AI assistant attached, see the comparison of free LinkedIn automation tools (/articles/completely-free-linkedin-automation-tool).

## Frequently asked questions

### Is there an official LinkedIn MCP server?

No. As of 6 November 2026, LinkedIn has not published an MCP server, and none is listed under a LinkedIn or Microsoft namespace in the official MCP Registry. Every LinkedIn MCP server is built by a third party.

### What is the best LinkedIn MCP server?

It depends on the job. For posting, an official-API connector such as Composio, Pipedream or Zapier. For open-source reading and search, stickerdaniel/linkedin-mcp-server. For outreach campaigns, a hosted workspace such as OutreachGenie. No server does everything without trade-offs.

### Is there a LinkedIn MCP server on GitHub?

Yes. stickerdaniel/linkedin-mcp-server (Apache-2.0, 4,119 stars) and eliasbiondo/linkedin-mcp-server (MIT, 212 stars) are open source and self-hosted. OutreachGenie's GitHub repository holds documentation and an audit skill; its hosted server code is not public.

### Does a LinkedIn MCP server need my cookie?

Session-based servers such as stickerdaniel/linkedin-mcp-server sign in as you and keep your LinkedIn session in a local browser profile. Official-API connectors use OAuth instead. Never paste a cookie or password into a chat, and never use someone else's session: LinkedIn's User Agreement prohibits copying cookies to use another person's account.

### Can a LinkedIn MCP server send messages or connection requests?

Only routes outside LinkedIn's open API can: session-based servers, unofficial APIs such as Unipile, and extension-based products such as OutreachGenie. Official-API connectors cannot. All of these sending routes carry account-restriction risk under LinkedIn's rules.

### Which LinkedIn MCP server works with Claude?

Remote servers such as Composio, Pipedream, Zapier, Apify, Unipile and OutreachGenie are added in Claude as custom connectors; Anthropic limits Free plans to one. Local servers such as stickerdaniel/linkedin-mcp-server run on your computer through Claude Desktop or Claude Code. The Claude and ChatGPT setup guide (/articles/connect-claude-chatgpt-to-linkedin) covers each route.

## Sources and further reading

- [Getting access to LinkedIn APIs](https://learn.microsoft.com/en-us/linkedin/shared/authentication/getting-access), LinkedIn on Microsoft Learn
- [User Agreement, section 8.2](https://www.linkedin.com/legal/user-agreement), LinkedIn
- [API Terms of Use](https://www.linkedin.com/legal/l/api-terms-of-use), LinkedIn
- [Prohibited software and extensions](https://www.linkedin.com/help/linkedin/answer/a1341387), LinkedIn Help
- [Official MCP Registry](https://registry.modelcontextprotocol.io/), Model Context Protocol
- [LinkedIn toolkit](https://docs.composio.dev/toolkits/linkedin), Composio
- [LinkedIn components](https://github.com/PipedreamHQ/pipedream/tree/master/components/linkedin), Pipedream on GitHub
- [MCP servers](https://pipedream.com/docs/connect/mcp), Pipedream
- [LinkedIn integrations](https://zapier.com/apps/linkedin/integrations), Zapier
- [Zapier MCP usage and billing](https://docs.zapier.com/mcp/overview/usage), Zapier
- [stickerdaniel/linkedin-mcp-server](https://github.com/stickerdaniel/linkedin-mcp-server), GitHub
- [eliasbiondo/linkedin-mcp-server](https://github.com/eliasbiondo/linkedin-mcp-server), GitHub
- [LinkedIn MCP server](https://apify.com/mcp/linkedin-mcp-server), Apify
- [Unipile MCP server](https://github.com/unipile/unipile-mcp), Unipile on GitHub
- [Unipile API pricing](https://www.unipile.com/pricing-api/), Unipile
- [LinkedIn MCP server for Claude and ChatGPT](https://www.outreachgenie.ai/linkedin-mcp), OutreachGenie
- [OutreachGenie LinkedIn MCP repository](https://github.com/Outreachgenie/outreachgenie-linkedin-mcp), OutreachGenie on GitHub
- [Pricing and plan allowances](https://www.outreachgenie.ai/pricing), OutreachGenie

Disclosure: we also make OutreachGenie.

Canonical URL: https://innovate-blog.com/articles/linkedin-mcp-servers-compared
