Grapevine MCP connector for Claude and ChatGPT

Yes, Grapevine has an MCP connector. If you are a Grapevine client, you can add it to Claude, ChatGPT or any other assistant that takes custom connectors, and then read your content plan, check what we know about you against what your assistant knows, tell us what's new, fix the text of a post and approve it from the chat you already have open. This page covers who it is for, how to connect it, eight requests to try, what your assistant cannot do, and how to disconnect it.

What the Grapevine MCP connector is

MCP stands for Model Context Protocol, an open standard that lets an AI assistant work with an outside service on your behalf. A connector is that service's side of it: one address you give your assistant, and a sign-in where you say yes.

The Grapevine connector lets your assistant into your own Grapevine account and nowhere else. Through it, your assistant can:

• read your content plan day by day, with the post written for each day and its status
• read one post in full, including every part of a thread and the choices of a poll
• read how your account is doing: followers, and what your posts and replies gained
• read your brief: what we have on file about you and write your posts from, with every empty answer named
• see the questions your plan has for you, and the facts already on record
• see what your plan is missing right now: a day waiting for a fact, an offer about to end, a number to confirm
• tell Grapevine something new in plain words: a release, a date, a number, an offer and its last day
• answer one of your plan's questions
• replace the text of a post that has not been published yet
• approve a post so it goes out on its day

It is the same account you see in the console, read and changed with the same rights you have there. Nothing in Grapevine requires it: everything above also works in the console, and the connector is there for the days when asking is faster than opening a tab.

Who it is for

The connector is for Grapevine clients: a project team or a founder with their own Grapevine login. It suits you if you already spend your working day in an assistant and want your content plan to be one question away, for example while you are writing release notes and realize the date belongs in next week's posts.

You need two things:

• a Grapevine client account
• an assistant that can add a custom MCP connector. Whether yours can may depend on the plan you have with that assistant, so check there first.

How to connect it

The address of the connector is:

https://gateway.grapevine.run/mcp

You will also find it in your Grapevine console, on the Connector screen, with a button to copy it. The same screen lists the assistants you have connected.

Assistants name their menus differently and move them between releases, so the steps below describe what to look for rather than what to click:

1. In your assistant's settings, open its connectors. Some assistants call them integrations, apps or MCP servers. Choose to add a custom one.
2. Paste the address above as the connector's URL. If it asks for a name, call it Grapevine.
3. Your assistant sends you to a Grapevine page. Sign in with your Grapevine login.
4. The page shows which app is asking, what it will be able to do and where your answer goes. If that is the assistant you were just using, press Allow.

After that, Grapevine appears among your assistant's connectors and you can ask it things in a normal chat.

Two things worth knowing if a step does not go through:

• The Grapevine page keeps a request open for ten minutes. If it says the request has expired, go back to your assistant and connect again.
• If the page says the connector is for client accounts, you are signed in with a login that has no content plan of its own. Sign in with the login you use for your console.

Eight requests to try

You do not need special commands. Say it the way you would say it to a colleague.

See your plan for the week

What is on my Grapevine content plan this week?

Your assistant lists the days, the topic of each, and whether its post is written, waiting for your approval or already out.

Read tomorrow's post

Show me the post that goes out tomorrow.

You get the full text. For a thread you get every part in order, and for a poll the question and its choices.

Add a fact with a date

Tell Grapevine that our mainnet launch moved to November 4.

The fact is on record right away, the same as when you type it into What's new in the console. The posts built on it show up on your plan a little later.

Answer a question from your plan

Does my plan have any questions for me? Answer the one about the offer: it still ends on the 30th.

Your plan sometimes needs something only you know: a number to confirm, whether an offer is still on, what is coming next month. Your assistant shows the open questions and passes your answer back.

Compare your brief with what your assistant knows

Read my Grapevine brief and compare it with what you know about our project. What is missing or out of date?

Your assistant reads what we have on file about you: what you do, who it is for, how you sound, your offers, numbers and dates. It tells you where that differs from what it knows. Nothing is added at this point. When you say which of it is right, your assistant passes on those facts, a few sentences at a time.

Keep it to what you would say in public. Do not ask your assistant to move a whole internal document into Grapevine: what you add here is what your posts are written from.

Ask what your plan is missing

What is my Grapevine plan missing right now?

You get a short list: a day that waits for something real from you, an offer close to its last day, a number nobody has confirmed in a month, an empty part of your brief. Each line comes with the way to close it, and your assistant asks you for the answer instead of guessing it.

Fix the text of a post

In Thursday's post, replace "soon" with "on November 4" and cut the last sentence.

Your assistant replaces the text of the post. This works for any post that has not been published yet.

Approve a post

I have read tomorrow's post. Approve it.

The post is approved and goes out on its day from your X account. Ask to see a post before you approve it: approving is the one request here that ends in something public.

What your assistant cannot do, and why

Your assistant cannot:

• delete a post or anything else
• ask for a post to be written again from scratch, or move a day on your plan
• read or change your replies or your reply plan
• change your brief: what you do, who it is for, your positioning, your voice. It can read all of it and add facts to it, and the rest you change in Settings
• change your settings: your project's name, your time zone, your connected X account
• change your plan with us or your billing
• reach anything outside your own account

The reason is the same for all of them. An assistant acts on a sentence, and a sentence can be misread. So the connector covers reading, and the few changes that are easy to see and easy to put right: a fact, an answer, the wording of a post that is not out yet, an approval. Anything that removes work, moves your month around or touches how you pay stays in the console, where you do it with your own hand and see the result on the screen in front of you.

What your assistant adds is marked as coming from your assistant in the Recent changes of your What's new screen, so you can always tell it from what you typed yourself.

Until you have chosen a Grapevine plan, your assistant can read everything in your account and change nothing.

One more thing to keep in mind: what your assistant reads from Grapevine passes through that assistant, under its own terms. It is the same text you would see in your console.

How to disconnect it

There are two ways, and either is enough:

• In your Grapevine console, open Connector, find the assistant under Connected assistants and press Disconnect.
• In your assistant's settings, find Grapevine among its connectors and remove it.

From then on that assistant can no longer read or change anything in your account. Your content plan, your posts and everything you told us through the assistant stay exactly as they are.

You can connect it again at any time with the same address.

FAQ

Does Grapevine have an MCP server?

Yes. Grapevine has an MCP connector for its clients at https://gateway.grapevine.run/mcp. Add it as a custom connector in Claude, ChatGPT or another assistant that supports MCP, then sign in with your Grapevine login and press Allow.

Which assistants does the Grapevine connector work with?

Any assistant that can add a custom MCP connector by its address and sign you in through a browser page, which includes Claude and ChatGPT. Whether custom connectors are available to you can depend on your plan with that assistant.

Can my assistant publish a post on its own?

No. It can approve a post when you ask it to, and an approved post goes out on its day. It cannot post anything outside your content plan, and it cannot publish at once.

Can my assistant delete posts or change my billing?

No. The connector has no way to delete anything, change your settings or touch your billing. Those stay in your Grapevine console.

Do I need the connector to use Grapevine?

No. Everything it does is also in your console. The connector is a second way in for people who already work inside an assistant.

How do I remove the Grapevine connector?

Press Disconnect next to the assistant on the Connector screen of your Grapevine console, or remove Grapevine from the connectors in your assistant's settings. That assistant then has no access to your account, and nothing in your content plan changes.