Narrative

Why I Cancelled My Cursor Subscription.
And what I use now.

This isn't a hit piece. Cursor is a good product made by good engineers. But I stopped paying for it, and the reason might matter if you're on the fence.

The weekend I noticed

Last summer, I burned through a month of Cursor credits in a single weekend. I wasn't even trying. I was building a small side project — typical stuff, backend endpoints, some UI — and at some point on Sunday evening the tool started reminding me I was running low.

I'd been paying $20 a month for Cursor Pro since the beginning. It hadn't occurred to me until that moment that I had no idea how much I was actually getting for it. The credit balance had always just been there, quietly ticking down, and I'd never had a reason to look at it.

I looked. I did some math. I felt something I didn't like.

The tool I was paying to save me time had quietly started charging me for thinking harder.

The three-card problem

I pulled up my last 12 months of subscriptions. Cursor Pro. Claude Pro. ChatGPT Plus. GitHub Copilot. Two different cards. Three different billing dates. Somewhere north of $800 for the year.

Each one was defensible on its own. Twenty bucks for a tool I used every day. Ten bucks for another one I used every day. The problem wasn't any single subscription. It was that I'd never sat down and looked at them together.

There's a quote floating around indie dev forums that stuck with me:

“It's just — everything is a subscription now, even the GPS. But if it was a one-time deal, I'd totally be up for paying.”

That was me. I'm not cheap. I'm tired.

What finally moved me off

It wasn't the pricing alone. It was the realization that Cursor and I had drifted into different use cases.

Cursor is an IDE-first tool. You sit in the editor, you chat with it, you accept edits. That's great when you're heads-down on one thing. I'm rarely heads-down on one thing. I have a full-time job and a list of side projects I want to ship in the margins of my week.

What I actually needed wasn't a faster editor. It was a tool that would work on my backlog while I wasn't at my desk. Something I could hand a sprint to, walk away from, and find the work done when I came back. Cursor wasn't built for that, and trying to force it was the real source of my credit burn.

What I use now

I built DevboardAI. (Full disclosure, obviously.)

It sits on top of the Claude Code, Codex, or Kimi CLI agents — whichever you have installed — and gives them a Kanban board. You describe a sprint, it generates the tasks, and the orchestrator runs them in parallel against your repo. Locally. On your Mac.

The part that matters for this post is the pricing: $74 once, never again. That number was deliberate. I wanted something I could recommend without caveating about seat math and credit rules. If you keep it for longer than three months, it's already cheaper than most of the stack I was on.

When to stay on Cursor

I'm not telling you to cancel. If you love the inline chat, if your workflow is genuinely editor-first, if you're on a team that uses Cursor as the standard — stay. It's a good product.

I'd tell you to cancel if: you've been surprised by your credit usage, you have more ideas than hours, or you're realizing you're paying for three tools that all do the same thing. In that last case, you're probably a month away from auditing your stack anyway. I'd just do it now.

Stop renting your agents.

Own the orchestrator instead. $74 once.