Tool · AI

Cursor

Our pick for now
Built on VS Code
Agents in the cloud

The editor we build in and the best place we've found to run Claude Code. Its own agents take the clear-cut coding jobs, on models that come with the plan.

Talk numbers All tools

21MCP servers connected (21 October 2026)
14repos it checks every month
7builds on it
8tools it connects to

Why we use it

Why it's our pick for now

Code, AI and terminal in one window
It's built on VS Code, so extensions and shortcuts carry over, and you can see what the AI's changing while it changes it.
Agents that work without you
Cloud agents work on their own branch on Cursor's servers, so a job keeps going after you close the laptop. We can watch one live and redirect it mid-run.
Its own models, in the plan
Cursor's models, like Auto and Composer, come with the subscription. They do the bulk of our coding jobs, and we keep Claude for the thinking.
Rules the whole team shares
Project rules live in the repository, so every person and every agent follows the same house style and safety rules.
Connected to your tools
Through MCP it reads analytics, Tag Manager, Google Ads, Search Console, our CRM, Webflow and GitHub, so an audit happens right where the code is.
Syncs with GitHub
Work happens on branches and pull requests, so a person reads every change before it's merged.

Our verdict

Cursor: our current call

Pick

The best place we've found to run Claude Code: it's an editor, it syncs with GitHub and our plan covers plenty of use.

Last known verdict

Reviewed 25 days ago

Why

It's an editor, so we can see the code and the AI working on it in one place.
It syncs with GitHub, so we can read every change as a pull request before it's merged.
Our plan includes enough use of Cursor's own models that bulk jobs don't need rationing.

What would change our mind

A better place to run Claude Code, or a plan change that makes Cursor a poor fit.

Cursor's the editor we build in, and its agents take our clear-cut coding jobs. We reviewed it on 5 October 2026 and it's our pick for now: it's an IDE, it syncs with GitHub, the base allowance is good, and it's the best place we've found to run Claude Code.

How the work is split

Claude Code plans, briefs and reviews. It works out the approach, writes a brief that says what "done" means, and reads what comes back.
Cursor's agents build the clear-cut parts: fixes, small features, refactors and docs. They run in Cursor's cloud or as a job on the laptop, each on its own branch.
Someone's always watching. We watch running agents and steer them mid-run. If one drifts, we tell it what to change instead of starting again, and a job that needs more than a couple of corrections moves to Claude.

Cursor's agents run on the models in Cursor (Composer, Auto or Grok), not Claude. We keep Claude for Claude Code. Which model does what is on the Anthropic page.

Jobs it runs on a schedule

Every month, a Cursor agent checks the docs in 14 of our repositories against their code, fixes anything that's drifted and opens a pull request. Claude Code reviews those and merges them. The one on our own admin repository waits for a person.

What's set up in it

One set of rules for two tools. Cursor and Claude Code follow the same house rules, and a check catches any gap. We built that: Cursor and Claude Code parity.
Connected to our other tools (MCP): analytics, Tag Manager, Google Ads, Search Console, our CRM, Google Docs, Webflow, GitHub, meeting transcripts and our finance system.
Grok Bot, Cursor's always-on agents: a small team of bots that works while the laptop's shut. They answer our questions, file call notes into the client record, check the sites and watch for failed builds. Anything that matters gets filed in the repository or the CRM, not left in chat.
Safety rails: nothing gets emailed, posted or published unless a person asks for it there and then. Jobs on the laptop can't open a browser or read its saved logins, and a check stops passwords and keys being saved into the code. AI builds, AI reviews and the checks run, but a person decides what goes live.

As of 21 October 2026: 21 tools connected to the editor through MCP.

What we pay for

We pay for Cursor's top plan, Ultra. Cursor's pricing page shows the other plans. Our scheduled docs checks run on Cursor's Auto model, which Ultra includes, and the plan lets us fan several agents out at once to build agentic harnesses: each takes its own part of a job and a person reviews what comes back. A subscription is a good way to manage cost, as long as your workspace lets several agents run at once without losing context.

What we've built with it

TechStack Analyser, the Markdown viewer and our docs hub were all built in Cursor, with Claude Code alongside, on our own rule set. This website's rebuild from Webflow runs on the same rules, with Cursor's agents and Claude Code splitting the coding jobs as above. We also run tracking and Tag Manager audits through the MCP connections (see technical tracking), and keep client details out of anything we share. The first big test of the Webflow connection is in How We Built 136 FAQs in Minutes.

There's a wider picture on our AI page, and we set up the same thing for other teams: AI setup.

Compared

Cursor or Claude Code?

We pay for both and use them together. Pick the job and we'll tell you which one gets it.

What's the job?

Clear-cut coding jobs

What matters	CursorOur pick	Claude Code
Examples	A bug fix, a small feature, a refactor, a docs update	The same jobs, using Claude allowance we'd rather keep for harder work
Models	Auto, Composer or Grok, never Claude	Sonnet by default, Opus rarely, on our own Claude plan
Where it runs	Cursor's cloud or a job on the laptop, each on its own branch	Your terminal, editor, the desktop app or the browser
Long runs	Cloud agents carry on after you close the laptop	Web sessions and scheduled routines do too

Our pick: Cursor. Fixes, small features, refactors and docs go to a Cursor agent on Auto, Composer or Grok. Claude Code writes the brief and checks the result.

Planning and judgement calls

What matters	Cursor	Claude CodeOur pick
Best at	Doing a clear brief quickly	Working out the approach, then building and checking it
Big jobs	One agent per job, each on its own branch	A lead session splits the job across sub-agents and merges what they send back
In our setup	Takes briefs and works on them	Writes the briefs, watches the agents and reviews their pull requests
Models	Auto, Composer or Grok	Sonnet by default, Opus only when a job really needs it

Our pick: Claude Code. Working out an approach, design-heavy pages and anything with open questions go to Claude. It also reviews what Cursor's agents send back.

Working in the code yourself

What matters	CursorOur pick	Claude Code
What it is	A code editor built on VS Code	An agent for your terminal, an editor panel, the desktop app or the browser
How you work	You edit, and the AI helps in the same window	It makes the changes and you review the diff
GitHub	Branches and pull requests from the editor	Commits, branches and pull requests on its own
Using both	Claude Code installs as an extension, so both share one window	It reads the same rules, generated from one source

Our pick: Cursor. It's a full editor, so you see the code, the AI and the terminal in one window, and Claude Code runs as a panel inside it.

Facts are from Cursor's and Anthropic's own pricing pages and docs, October 2026, and from how we run both today. Plan prices are in the FAQ below; the rest is from Cursor's docs, October 2026.

Wires

What we connect it to

On its own

AI-written code still needs a person to read it, test it and own it. If nobody checks, a faster editor just means faster mistakes.

Show 4 more

GTM

GA4

Google Ads

Proof

Where we've used Cursor

All our work

Services

Services that use it

We don't sell it on its own. Each of these services uses it alongside the rest of your setup.

How we start

About the tool

What Cursor is

Cursor is a code editor, built on VS Code, with AI models inside it. You ask in plain English and it reads, writes and changes code across a whole project. Its agents can also run in Cursor's cloud, on their own branch, and come back with a pull request.

Questions

Questions about Cursor

Straight answers on how we work, what it costs and what to expect. If yours isn't here, just ask.

Still have questions?

Talk to us

What does Cursor cost and which plan should you choose?

Cursor has a free plan and several paid ones, and what separates them is how much AI usage each includes before you hit a limit. Start on the lowest plan that covers what you do every day and move up when you hit a limit. We're on the top plan, Ultra. Cursor renames and reprices its plans, so check Cursor's pricing page for current figures.

Read the full answer : What does Cursor cost and which plan should you choose?

How does Cursor handle code security and privacy concerns?

Cursor has a Privacy Mode and says that when it's on, it won't train on your data. It also says it holds a SOC 2 Type II attestation. Your code still goes to AI model providers to get answers, so the settings, the secrets and the review process around it matter too.

Read the full answer : How does Cursor handle code security and privacy concerns?

What are the best practices for getting the most from Cursor's AI features?

Give it a clear brief that says what "done" means, keep your rules in the project so each agent follows them, and put each job on its own branch. Then read the result before you merge it.

For bigger jobs, ask for a short design note first and check it before any code is written.

Read the full answer : What are the best practices for getting the most from Cursor's AI features?

How does Cursor's AI assistance improve development productivity?

You describe a change in plain English and Cursor reads the project, makes the edit across the files and shows you the result. Its agents can also work on their own branch while you do something else.

It only saves time if a person reads what comes back. We review each change before it ships, and we haven't measured the time saved.

Read the full answer : How does Cursor's AI assistance improve development productivity?

What programming languages and frameworks does Cursor support?

Cursor is built on VS Code, so it works with the same languages and frameworks that VS Code does, through its extensions. The AI works across whatever is in your project.

We use it for TypeScript, JavaScript, Astro, React and Python.

Read the full answer : What programming languages and frameworks does Cursor support?

What platforms do you use for website development?

We pick the platform for the site, not the other way round. Webflow suits a site your team edits itself, Astro with Keystatic suits a site that needs custom code (our own site runs on it), and for shops we use Shopify (we've built a custom Webflow storefront on Shopify).

We don't build or maintain WordPress. If you're on it, we'll move you to Webflow or Astro, whichever suits the site.

Read the full answer : What platforms do you use for website development?

How do you integrate Cursor with project management and development workflows?

Cursor connects to GitHub, so work happens on a branch and arrives as a pull request that someone reads before it merges. Our agents can also reach tools like Tag Manager, Google Ads and our CRM through MCP connections, including filing tasks in it.

We use Cursor with GitHub, so every change can be reviewed, tracked and undone.

Read the full answer : How do you integrate Cursor with project management and development workflows?

How do you use Cursor for building marketing automation tools?

Our marketing automation runs in n8n. Cursor is the editor we build in, so it's where we write and review the code around it: small scripts, integrations and tools that connect your forms, sheets and CRM. A person checks everything before it goes live.

Read the full answer : How do you use Cursor for building marketing automation tools?

What are the key differences between Cursor and VS Code?

Cursor is built on VS Code, so it looks and feels the same and your extensions and shortcuts carry over. The difference is that AI is part of the editor: it can change code across a project, and its agents can work on their own branch, in Cursor's cloud or on your own machine.

We build in Cursor because it syncs with GitHub and is the best place we've found to run Claude Code.

Read the full answer : What are the key differences between Cursor and VS Code?

Why do you run Claude Code inside Cursor?

Cursor's an editor with GitHub sync and a good base allowance, so the code, the agent and the terminal all sit in one window. It's where Claude Code works best for us so far. That's our pick for now, and we'll say if it changes.

Show 2 more questions

Why pay for Claude as well as Cursor?

Because they do different jobs, and some days we hit the limits on one of the tools we pay for, so our work shouldn't depend on just one. Cursor's agents run Auto, Composer or Grok models on the Cursor plan, and Claude runs through Claude Code on our Claude plan, so each plan covers its own work and neither has to carry the other. The right plans depend on the person and what they do.

Should I use Cursor or Claude Code?

If you write code yourself, start with Cursor. It's a full editor, so you see the code, the AI and the terminal in one window. If you want an AI to take a job from plan to pull request, add Claude Code. It runs inside Cursor, so you don't have to choose.

We use both. Claude Code plans the work, builds the parts that need judgement and reviews the changes before they merge. Cursor's agents take the clear-cut jobs, like fixes, small features and docs, on Cursor's own models that come with the plan (check Cursor's pricing page).