TL;DR: Teamfluence has a remote MCP server at https://api.teamfluence.com/mcp. Paste that URL into Claude, ChatGPT or Grok, sign in with your Teamfluence account, and you can ask questions about who engaged with your team on LinkedIn in plain language. No API key. No terminal. No JSON file to edit.
The LinkedIn engagement MCP is a Model Context Protocol server that gives an AI assistant secure, read-and-update access to the LinkedIn engagement signals Teamfluence has already collected for your workspace.
What problem does this actually solve?
Collecting the signals was never the hard part. The hard part is the question you ask of them on a Tuesday morning when you have twenty minutes and a pipeline review at eleven.
"Which of the people who viewed my profile this month work at companies over 200 people?"
That question has always been answerable. It just meant opening a dashboard, remembering which filter did what, and clicking through four screens to get there. Most people did not bother. The signals sat in the tool and the rep sent the same generic follow-up they sent last week.
Now you type the question into the assistant you already have open.
What is MCP, in one paragraph?
Model Context Protocol is an open standard for connecting AI assistants to outside systems. Think of it as a common plug. Before MCP, every tool needed its own custom integration with every assistant. With MCP, a tool exposes a set of capabilities once, and any assistant that speaks the protocol can use them. Claude, ChatGPT and Grok all speak it.
How is this different from every other sales MCP?
Most MCP servers in the sales tooling category follow the same pattern: you generate an API key, install a package from npm, then hand-edit a JSON config file to point your assistant at a server running on your own machine. That works, and it is genuinely fine if you are technical. It also stops roughly everyone else at step one.
Teamfluence went the other way.
| The common pattern | Teamfluence MCP | |
|---|---|---|
| Where it runs | Locally, on your machine | Hosted by us, always on |
| Setup | Install a package, edit a config file | Paste one URL |
| Auth | Generate and paste an API key | Sign in with OAuth, like any app |
| Key handling | You store a secret in a text file | No key exists to leak |
| Works in | Usually Claude Desktop only | Claude, ChatGPT and Grok |
| Access control | Tied to whoever holds the key | Tied to your Teamfluence login and permissions |
The auth difference matters more than it looks. An API key in a config file is a long-lived secret sitting in plain text on a laptop. Our server uses OAuth 2.1 with PKCE and dynamic client registration, which is the same class of sign-in flow you already use when you connect Slack to something. You approve access in a browser, you can revoke it later, and there is no secret for anyone to copy out of a file.
For agencies, that is the part worth reading twice. Onboarding a freelancer to a client workspace means they sign in and get exactly the access their account allows. Offboarding means revoking it. Nobody is passing API keys around in Slack.
What can it actually do?
Three tools, and being straight about the boundaries is worth more to you than a longer list:
describe_workspacegives the assistant the shape of your workspace so its answers are grounded in your setup rather than guessed.fetch_leadspulls the leads and engagement signals Teamfluence has collected.update_leadwrites back. Status changes, qualification, the housekeeping that usually happens one click at a time.
What it does not do, on purpose: it does not post to LinkedIn, send connection requests, or write DMs on your behalf. This is a server for reading your pipeline and keeping it tidy, not for automating outreach out of a chat window. If you were hoping to point an agent at LinkedIn and let it run, this is not that, and we would rather tell you now.
Things worth asking once it is connected:
- "Who engaged with our team's posts more than twice in the last 30 days but is not in our CRM yet?"
- "Show me profile views from companies over 200 employees, sorted by how recently they looked."
- "Which of my tracked contacts went quiet after engaging three or more times?"
- "Summarise this week's signals by company and tell me which three accounts are heating up."
- "Mark everyone from that list as qualified."
That last one is a write, so your assistant will ask you to confirm before it changes anything.
How do I connect it to Claude?
You need a Teamfluence account. Custom connectors work on Claude Free, Pro, Max, Team and Enterprise, though Free is limited to one.
- In Claude, go to Customize > Connectors.
- Click + and choose Add custom connector.
- Paste
https://api.teamfluence.com/mcpand click Add. - Sign in with your Teamfluence account when the browser window appears.
- In any chat, click +, choose Connectors, and switch Teamfluence on.
On a Team or Enterprise plan, an Owner adds the connector once under Organization settings > Connectors, and everyone else clicks Connect on it from their own Customize > Connectors screen.
How do I connect it to ChatGPT?
ChatGPT needs Developer Mode switched on first. It is available on Plus, Pro, Business, Enterprise and Edu, on the web.
- Go to Settings > Security and login and turn on Developer mode.
- In Settings > Connectors, add a custom connector pointing at
https://api.teamfluence.com/mcp. - Choose OAuth when asked how to authenticate, then sign in to Teamfluence.
- Start a chat, click +, then More, then Developer Mode, and pick Teamfluence.
ChatGPT asks you to confirm write actions before it runs them. When you ask it to update a lead, expect a prompt. Read what it plans to change before you approve it.
How do I connect it to Grok?
Grok added support for custom MCP servers in May 2026.
- Go to grok.com/connectors.
- Choose New Connector, then Custom.
- Paste
https://api.teamfluence.com/mcpand complete the sign-in.
One caveat. xAI's documentation lists custom MCP connectors as a Grok Business and Enterprise feature, and a workspace admin may need to approve the connector before your team can use it. If you do not see the option, that is usually why.
Does this replace the dashboard?
No, and it would be a strange thing to want. The dashboard is better when you want to see everything at once and browse. The MCP is better when you have one specific question and no appetite for clicking.
Where it genuinely changes the work is the boring middle: the weekly pipeline review, the "did anyone from this account engage recently" check before a call, the Friday tidy-up of lead statuses. Those are the tasks people skip when they take four minutes each. They stop getting skipped when they take fifteen seconds.
Is my data safe?
The server only reaches data your Teamfluence account can already reach. Sign-in runs through OAuth, so your assistant receives a scoped token rather than a password or a permanent key, and you can revoke that access whenever you want without rotating anything for the rest of your team.
Worth knowing regardless of vendor: when you connect any MCP server to an assistant, the data that server returns becomes part of that conversation. Apply the same judgement you would apply to pasting a lead export into a chat window. Anthropic, OpenAI and xAI all publish their own data handling terms, and which assistant you choose is a decision about whose terms you are comfortable with.
Frequently asked questions
Do I need an API key to use the Teamfluence MCP?
No. The server uses OAuth, so you sign in with your normal Teamfluence account. There is no key to generate, store or rotate.
Which AI assistants support the Teamfluence MCP?
Claude, ChatGPT and Grok. Any assistant that supports remote MCP servers over HTTPS should work, since the server follows the open standard rather than anything proprietary.
Does the MCP server work with the free trial?
Yes. Teamfluence trials run 7 days with no credit card, and the MCP is available during the trial.
Can the MCP send LinkedIn messages or connection requests?
No. The current tools read your workspace and update leads. Sending is not exposed through MCP.
Is there an extra cost for the MCP server?
No. It is part of your Teamfluence workspace.
What happens if I revoke access?
The assistant loses its token and can no longer reach your workspace. Your data and your Teamfluence account are unaffected.
Try it
If you already use Teamfluence, the connector works right now. Paste https://api.teamfluence.com/mcp into Claude, ChatGPT or Grok and ask it who has been paying attention to your team this month.
If you do not, start a free trial. 7 days, no credit card. Connect your team's LinkedIn activity, let Teamfluence sort the 15.6% that match your ICP from the rest, then ask your assistant what to do about it.