Best Cursor Data Connectors in 2026: Plugins and MCP Servers for Marketing Data

Cursor is a code editor, but a lot of marketing work now happens in it: SQL for attribution models, scripts that clean campaign exports, internal reporting apps. Cursor AI already reads your codebase. It writes better code and gives better answers when it can also see the data that code works with. Cursor data connectors give it that access, so you can check last month's ad spend in the same window where you're editing the report that uses it.

This guide compares the Cursor data connectors that matter for marketing teams: official plugins in the Cursor Marketplace, multi-source platforms that fill their gaps, and Google's own servers for Google data. For each, you'll find the data covered, how it connects, what it costs and what to watch for.

Editorial Team Nov 3, 2026 · 13 min read

How Cursor MCP servers work

Cursor connects to external tools through the Model Context Protocol (MCP), an open standard for linking AI agents to data and apps. The Model Context Protocol is the same one used by Claude Code, Claude and ChatGPT. Cursor supports MCP servers in the editor and in Cloud Agents, and the setup is the same for Cursor IDE MCP servers and the ones Cloud Agents use.

Each server exposes a set of MCP tools, such as "run a report" or "list ad accounts", and the agent picks which tool calls to make based on your request. By default, Cursor asks you to approve each call and shows the arguments it plans to send.

Servers connect in one of two ways:

Local (stdio). The server runs on your machine, usually launched with a command like npx for Node.js packages or pipx for Python. You handle credentials yourself, typically through environment variables. A local server can also reach your filesystem, so run only ones you trust.
Remote (SSE or Streamable HTTP). The server runs on the vendor's side, and you sign in with OAuth. Remote servers are easier to set up and share across a team.

The Cursor MCP servers configuration lives in an mcp.json file. Put it in .cursor/mcp.json inside a project to share it with everyone who works in that repo, or in ~/.cursor/mcp.json to make the servers available in every project. If both files define a server with the same name, the project file wins. Use ${env:NAME} references for API keys instead of pasting them into the file.

There are three places to find MCP servers for Cursor:

The Cursor Marketplace: official plugins with one-click install. A plugin can bundle an MCP server with skills, rules and hooks.
cursor.directory: community listings, many with an "Add to Cursor" button.
Vendor documentation: a server URL or install command that you add to mcp.json yourself.

One more difference affects your answers: where the data comes from at the moment you ask. Some connectors query the source API at request time. Others read a dataset that a pipeline refreshes on a schedule, so the answer reflects the last successful run. The first gives you the latest numbers but depends on the API's limits. The second gives you cleaned, combined data that you can trust to stay consistent between questions.

Types of Cursor data connectors

Cursor data connectors for marketing fall into three groups, depending on who builds them and how many sources they cover.

Type	Examples	Sources per server	Setup	Best for
Cursor Marketplace plugins	HubSpot, Semrush, Klaviyo, TikTok for Business	One platform each	One click from Cursor	Questions about one tool your team already pays for
Multi-source data platforms	Coupler.io, Windsor.ai, Supermetrics, Funnel	100–600+	Remote server, OAuth	Ad platforms the Marketplace doesn't cover, and cross-channel questions
Google's own servers	Google Analytics MCP, Google Ads MCP, BigQuery remote MCP	One platform each	mcp.json entry (BigQuery also one click), Google Cloud credentials	Free, direct access to Google data

Cursor connectors in the Marketplace also cover the developer tools your team may already use alongside marketing data: Figma, GitHub, the Slack MCP server, Google Drive, Cloudflare, and PostgreSQL hosts such as Neon Postgres and Supabase. These help with building and shipping, but they don't bring in campaign performance data.

Best MCP servers for Cursor and plugins compared

Here's how the best MCP servers for Cursor compare for marketing data, grouped by type. The table also works as a short Cursor MCP servers list. Marketplace plugins run on the vendor account you already have. Multi-source platforms charge separately, and their pricing models differ (accounts, data sources, flexpoints), so entry prices don't compare one to one.

Tool	Type	Data covered	Cost to use with Cursor	Best for
HubSpot	Marketplace plugin	CRM records, campaigns, content	Included with any HubSpot account	Campaign-to-pipeline questions
Salesforce	Marketplace plugin	CRM records, Flows, Data 360	Your Salesforce org (setup needed)	CRM reporting on Salesforce
Semrush	Marketplace plugin	Keywords, backlinks, traffic	SEO Classic Pro or Semrush One Starter and up, plus API units	SEO and competitor research
Ahrefs	Marketplace plugin	Keywords, backlinks, rankings	Lite plan or higher	SEO for existing Ahrefs users
Klaviyo	Marketplace plugin	Email campaigns, flows, profiles	Your Klaviyo account	Email and flow performance
TikTok for Business	Marketplace plugin	TikTok Ads	Your TikTok Ads account	TikTok campaign reporting
Coupler.io	Multi-source	400+ sources	Free plan; from $24/mo billed annually ($32 monthly)	Cleaned, combined cross-channel data
Windsor.ai	Multi-source	350+ sources	30-day trial, then Forever Free (1 source); from $19/mo billed annually ($23 monthly)	Wide coverage at a low price
Supermetrics	Multi-source	175 sources	14-day trial; from $44/mo billed annually ($55 monthly)	Teams already on Supermetrics
Funnel	Multi-source	117 to 600+, by plan	30-day trial; from $300/month billed annually	Large teams with a central data hub
Google Analytics MCP	Google's own server	GA4	Free, open source	Direct GA4 reports
Google Ads MCP	Google's own server	Google Ads	Free	Read-only Google Ads queries
BigQuery remote MCP	Google's own server (also in the Marketplace)	Your BigQuery tables	BigQuery usage pricing; sandbox available	Data already in a warehouse

Prices checked in October 2026.

The best Cursor data connectors for marketing teams, reviewed

This list is selective. It covers the connectors most useful for marketing data, starting with official Cursor Marketplace plugins and then the platforms that fill their gaps.

Cursor Marketplace plugins for marketing data

Marketplace plugins are official, install in one click from Cursor and use the account you already have with that vendor. Each one covers a single platform. These six are the most useful for marketing work.

HubSpot

HubSpot lets Cursor search and update contacts, companies, deals and tickets. HubSpot's remote server also reads campaigns, landing pages, blog posts and marketing events, and it's available to all HubSpot accounts.

Keep in mindIf sensitive data is turned on in your account, calls, emails, meetings, notes and tasks are blocked.

Best for: linking campaigns to pipeline and deals.

Salesforce

Salesforce queries, creates and updates records in your org. Its hosted servers follow each user's Salesforce permissions and can also run Flows and Data 360 queries.

Keep in mindThe hosted servers need setting up in your org before anyone can connect.

Best for: CRM reporting for teams on Salesforce.

Semrush

Semrush brings keyword, backlink, traffic and competitor data into Cursor.

Keep in mindYou need SEO Classic Pro or Guru, or Semrush One Starter or Pro+, and every request uses API units (50,000 included).

Best for: SEO research next to the content or site code you're working on.

Ahrefs

Ahrefs covers keywords, backlinks, rankings and site health.

Keep in mindIt needs a paid plan from Lite up, and MCP connections default to 200,000 API units a month.

Best for: teams that already use Ahrefs for SEO.

Klaviyo

Klaviyo reads campaign and flow performance, profiles and segments, and can create email campaigns.

Keep in mindYou need an Owner, Admin or Manager role. Turn on read-only mode if you don't want Cursor to change anything.

Best for: comparing email campaigns and flows by conversions.

TikTok for Business

TikTok for Business manages TikTok campaigns, ad groups, ads, audiences and creatives, and pulls reports.

Keep in mindIt covers TikTok only, and it can change live campaigns, so keep tool approval on.

Best for: TikTok Ads reporting and campaign changes.

Also in the Marketplace: Amplitude, Mixpanel and PostHog for product analytics, and Snowflake and Google Sheets for data you already store.

Plugins work well when one platform answers your question. As of October 2026, Google Analytics, Google Ads, Meta and LinkedIn don't have their own plugins in the Marketplace. Third-party plugins such as Adspirer cover several ad platforms, but they don't bring your CRM or analytics data in alongside them.

Multi-source platforms fill that gap: they connect your ad accounts, CRM and analytics through one server, and some clean and combine the data before Cursor sees it.

Coupler.io

Coupler.io is a data integration and AI analytics platform with 400+ sources. Its official server is on cursor.directory. You describe the data you need to Coupler AI, and a data flow sends it to Cursor, combining sources like Google Ads, Meta Ads and your CRM. When you ask a question in Cursor, Coupler.io's Analytical Engine calculates the figures from that data, and the model explains them.

Cursor is available as a destination on every plan, including Free. Paid plans start with Starter at $24/mo billed annually, with daily refresh. Pro refreshes hourly, and custom plans refresh every 15 minutes. Coupler.io also sends data to spreadsheets, warehouses, BI tools and other AI tools such as Claude and ChatGPT.

Keep in mindCursor reads the result of the last successful data flow run, so the refresh schedule decides how current your answers are. Chats that are already open may not pick up new data until you start a new one. The Free plan refreshes manually and returns up to 100 rows per run.

Best for: marketing teams that want cross-channel data cleaned and combined before the agent sees it.

Windsor.ai

Windsor.ai is a marketing data platform with 350+ sources, including ad platforms, analytics, ecommerce, CRMs and databases. Its hosted server supports Streamable HTTP and SSE, signs you in with OAuth, and also comes as a dedicated Cursor plugin. MCP access is included on every plan.

Basic costs $19/mo billed annually for 3 data sources with daily refresh. Faster refresh starts on Standard (hourly), and Professional goes down to 15 minutes.

Keep in mindLarge requests can fail with a size-limit error, so you may need to narrow date ranges or fields. Long pulls come back as pending jobs instead of immediate answers.

Best for: small teams that want broad source coverage at a low entry price.

Supermetrics

Supermetrics is a marketing data platform with 175 sources available through its MCP server, covering ads, analytics, email, CRM and SEO tools. Its documentation names Cursor as a compatible client. You connect with OAuth (recommended) or an API key. Beyond reading data, some beta tools can create and update campaigns, and new campaigns are always created as paused. Starter costs $44/mo billed annually for 3 data sources and 3 accounts per source.

Keep in mindThe MCP documentation says you need a subscription with API access, while the pricing page lists API access under Enterprise. Confirm which plan covers MCP before you buy. Account limits per source apply to MCP queries too.

Best for: teams already using Supermetrics for reporting that want the same data in Cursor.

Funnel

Funnel is a marketing data hub with 600+ connectors on Enterprise (117 on Starter). Funnel MCP is included on all plans and works with Cursor and other MCP clients. It reads the data you've already collected in Funnel, using your existing Funnel permissions. Starter starts at $300/month billed annually, priced through a mix of plan and flexpoints (purchased capacity).

Keep in mindThere's no free plan, and the entry price is the highest in this list. MMM and attribution results are only available if you also pay for Funnel Measure.

Best for: larger teams and agencies that already store their marketing data in Funnel.

Google's own MCP servers

Google Analytics and Google Ads aren't in the Marketplace, so Google publishes its own servers for them, which you add to mcp.json. BigQuery's server is also listed in the Marketplace as Google Cloud BigQuery.

Google Analytics MCP: official and marked experimental. It runs locally over stdio through pipx, needs Python 3.10+ and Google Cloud credentials, and offers read-only reports, including funnel and realtime reports.
Google Ads MCP: official and read-only. It runs locally, or remotely on Google Cloud Run with OAuth, and queries your accounts with GAQL.
BigQuery remote MCP: a Google-hosted server at bigquery.googleapis.com/mcp. It lists datasets and runs SQL, with results capped at 3,000 rows and queries canceled after three minutes by default.

Keep in mindThese are free, but setup needs a Google Cloud project, API access and IAM roles. They're the best Cursor MCP servers for direct Google data, and the least practical if no one on your team manages Google Cloud.

How to add MCP servers to Cursor

There are three ways to add a server. Here's how each works, with Coupler.io as the example for the directory route.

From the Cursor Marketplace. Open Customize > MCPs in Cursor, find the plugin and click Add to Cursor. Complete the sign-in if the plugin asks for one.

From cursor.directory (Coupler.io example):

Open the Coupler.io entry on cursor.directory and click Add to Cursor.
In Cursor, click Needs authentication next to the server and sign in with your Coupler.io account.
In Coupler.io, describe the data you need to Coupler AI or create a data flow with your sources, for example Google Ads and HubSpot.
Select Cursor as the destination and run the data flow. Set a schedule if you want the data refreshed.
Open the agent in Cursor and ask your question. Approve the tool call when Cursor asks.

By hand. Add the server to the mcp.json file in your project's .cursor folder or in the .cursor folder in your home directory. A remote server needs only its URL, for example https://mcp.windsor.ai/ for Windsor.ai. A local server needs the command that starts it, such as npx or pipx, plus any credentials it reads from environment variables.

Restart Cursor after editing the file, then check that the server shows as connected in your MCP settings.

A note on security. Cursor AI MCP servers can read, and sometimes change, data in your accounts. Their responses also go straight into the agent's context, which makes prompt injection a real risk with servers you don't trust. Install servers from vendors you know, keep tool approval on for anything that can write data, and use read-only scopes where the vendor offers them.

FAQ

Does the same setup work in Claude Code and other AI tools?

Mostly, yes. Remote servers follow the Model Context Protocol and use a URL and OAuth that any MCP client can use, so the same Coupler.io, Windsor.ai or Supermetrics account works in Claude Code, Claude or ChatGPT. Only the config format differs between clients and AI agents. Marketplace plugins are Cursor connectors, so they don't carry over.

Can Cursor change my ad campaigns through MCP?

It depends on the server. The Google Ads and Google Analytics servers are read-only. HubSpot can create and update CRM records, and Supermetrics has beta tools that create campaigns (as paused). Cursor asks for approval before each tool call by default, so check which MCP tools a server exposes and what they can write before you let them run automatically.

How do I share MCP servers with my team?

Commit .cursor/mcp.json to your repository so everyone working in it gets the same MCP servers in Cursor. Each person still signs in with their own account. On Teams and Enterprise plans, admins can add servers to a team marketplace, and Enterprise admins can limit which tools are allowed.

Where to start

The best MCP servers to use with Cursor are the ones that answer the questions you ask most often. Each of the Cursor MCP servers you add puts its tool definitions into the agent's context, so start with one or two. Developer tools like the GitHub MCP server, Figma or Cloudflare can come later, once your data access works. The right Cursor data connectors depend on where your data lives today:

If one platform answers most of your questions, such as HubSpot, Semrush or Klaviyo, install its plugin from the Cursor Marketplace.
If you need cross-channel numbers already cleaned and combined, start with Coupler.io. It joins and filters your sources first, so Cursor works from one combined dataset.
If you want a low entry price and pull data straight from the source, try Windsor.ai on its 30-day trial.
If your team already pays for Supermetrics or Funnel, use their MCP servers before adding another tool.
If your data already lives in BigQuery or you only need Google Analytics and Google Ads, Google's own servers are free, as long as someone on your team can set up Google Cloud credentials.

Written byMarketing Data Connectors Editorial Team

We write practical guides and side-by-side comparisons to help marketers choose the right tools for getting their data where they report.