MCP Servers: The Complete List of App Integrations (2026)
An MCP server is the piece that lets an AI client — Claude, ChatGPT, Cursor, and others — reach into one of your apps and read or write data through a shared interface. One app, one server. Below is the running list of apps that have an MCP server today, grouped by what they do, so you can jump straight to the one you need — plus the catch that applies to every server on the list, and what to use when you want the AI to actually run work instead of just answer questions about it.
If you’re new to the protocol itself, What Is MCP? covers it in a sentence and a diagram, and How to Install an MCP Server walks the setup. This page is the map: which apps you can connect, and where the whole approach stops.
What an MCP server actually gives you
Connected, an MCP server hands an AI client a set of tools scoped to one app — search, read, create, update. Ask a question in a chat and the AI answers grounded in your real data; tell it to make an edit and it makes the edit. Some apps ship an official, hosted server (Notion, Stripe, Linear, and a growing set maintain their own); others rely on a community or self-hosted server you run yourself. Each app’s page below says which kind it is and exactly which actions the server exposes.
The complete MCP server list, by category
Every link goes to a full breakdown of that app’s MCP server — what it does, whether it’s official, how to connect it, and where it stops.
CRM & Sales Salesforce · HubSpot · Pipedrive · Attio · Apollo · GoHighLevel · Gong
Email & Marketing Gmail · Outlook · Mailchimp · Klaviyo · ActiveCampaign · Instantly · Google Ads · Meta Ads · Semrush · Ahrefs · Customer.io · Resend · SendGrid
Productivity & Docs Notion · Airtable · Coda · Google Sheets · Google Docs · Google Slides · Confluence · TickTick
Project Management Asana · Monday · ClickUp · Trello · Basecamp · Wrike · Jira · Linear · Shortcut
Communication & Messaging Slack · Microsoft Teams · Discord · Telegram · WhatsApp · Zoom
Customer Support Zendesk · Freshdesk · Freshservice · Intercom · PagerDuty
Files & Storage Dropbox · Box · Google Drive · SharePoint
E-commerce & Payments Shopify · Stripe · Square · Wix · Webflow · PandaDoc · WooCommerce · Amazon
Accounting & Finance QuickBooks · Xero · YNAB · Brex
Analytics Google Analytics · Google Search Console · Apify
Social & Content LinkedIn · YouTube · Reddit · Canva · Adobe · Miro · Figma · Contentful · Instagram
Forms & HR Typeform · Rippling · Ashby
Suites Zoho
Working inside a specific AI client? The client-side connectors matter too: ChatGPT MCP and Carly’s MCP cover how each side speaks the protocol.
The one limit every MCP server shares
Pick any server on that list and the same four things are true. This isn’t a knock on any single app — it’s the shape of the protocol:
- It only works inside a chat you start. An MCP server waits for you to open a client and ask. Close the window and nothing happens — nothing is watching the app on your behalf.
- No triggers. A new deal in the CRM, an invoice going overdue, a support ticket landing — none of these can start anything. MCP has no “when this happens, do that.”
- One app at a time. The Notion server knows Notion; the Slack server knows Slack. Getting a new project to create a Slack post and a calendar event and a spreadsheet row means wiring up a separate server for each and hoping your client can juggle them in a single turn.
- You own the plumbing. Hosting (for community servers), OAuth scopes, and the blast radius of read/write access are all on you.
So MCP is a genuinely good way to ask an AI about your apps and make one-off edits from a chat. It is not a way to make those apps run — to have work happen on a schedule, or in reaction to something, across all the tools a task actually touches.
Connect AI to all of it — without running a single server
That “run on its own, across apps” gap is exactly where Carly fits. Carly connects natively to ~260 apps — including most of the ones on the list above — and to anything else with a public API through your own key, pasted on your integrations dashboard. There’s no server to host, no OAuth plumbing to maintain, and no client to keep open.
The difference that matters: Carly’s workflows are triggered and scheduled. The same connections an MCP server gives one app, Carly wires across many — and runs them whether or not a chat window is open:
- When a new deal is marked won in your CRM → create the onboarding doc, post it to the right Slack channel, and schedule the kickoff call — automatically, the moment it happens.
- Every Monday at 8am → pull last week’s numbers from Analytics and Stripe into a Sheet and email the summary — no one prompting it.
- When an invoice goes overdue in QuickBooks → draft the follow-up, log the task, and ping the account owner.
Those are the things MCP can’t reach: a trigger, a schedule, and more than one app in a single motion. Carly is free for unlimited Zapier-style workflows, with AI agents from $35/month — so the non-AI plumbing that stitches your apps together isn’t metered.
MCP is the right tool when you want to talk to an app from a chat. Carly is the right tool when you want the work to happen without you in the loop.
FAQ
What’s the difference between an MCP server and a connector? They’re the same idea from two sides. An MCP server exposes one app’s data and actions; an MCP client (the AI tool) connects to it. “MCP connector” is the everyday word for a server you’ve hooked up. Carly, by contrast, connects to apps directly rather than through per-app MCP servers.
Is there an official directory of MCP servers? There’s no single canonical registry — servers are scattered across vendor docs, GitHub, and community lists, which is why this page groups the ones that exist by category. Some apps (Notion, Stripe, Linear, and others) host and maintain their own official server; many rely on community-built ones.
Do I need to code to use an MCP server? For officially hosted servers, usually not — you add a remote URL in your AI client and sign in through OAuth. Community or self-hosted servers often require running the server yourself, which does involve some setup. How to Install an MCP Server covers both paths.
Can an MCP server trigger an automation? No. MCP servers respond to requests inside a chat; they can’t watch an app and start work when something changes. For triggered or scheduled work across apps, you need an automation layer like Carly rather than a raw MCP server.
Ready to automate your busywork?
Carly schedules, researches, and briefs you—so you can focus on what matters.
See what people say
"Before Carly, I relied on a Calendly link, but the whole process felt impersonal and not very professional. Carly changed that by handling all the back-and-forth, so I'm no longer stuck in endless email threads trying to line up schedules.
Now Carly reaches out to candidates, shares my real-time availability, lets them pick a slot, then sends a Zoom link and drops it straight into my calendar. She sends reminders to both of us before each call, which has significantly reduced no-shows and last-minute confusion.
On top of scheduling, Carly acts like a full executive assistant, sending me my schedule the night before so I can prepare for each call. It reminds me of the old x.ai assistant, but Carly is noticeably smarter, faster, and better suited to my healthcare recruitment business."
