Founder Playbook · Starter Story

8 tactics from Pontus

Cursor Directory$35K/month

My Website Makes $35K/Month (Built in 3 Hours)

Watch the full episode
Idea validation
I downloaded some cursor things and then I discovered this common pattern about rules so it was kind of mind-blowing for me that there were no place to find these rules in one specific place. You could Google and find some obscure gist on GitHub or some forums but no real clear answer.

Spot Directory Opportunities In Your Own Frustration When No Central Place Exists

Pontus spotted the opportunity by being a user himself, downloading tutorials on a flight and noticing rules kept appearing without a central home. The validation signal was his own frustration multiplied by the obscurity of existing answers (gists, forums) for a fast-growing tool. The market gap was already screaming.

Shipping
In a matter of minutes he spun up Figma and started to design while I started initial project with Next.js and I created a hard-coded JSON file with some rules that I found online and in matter of like an half an hour we had something up in Vercel ready to to click around on.

Hard-Code A JSON File And Push To Vercel Within Thirty Minutes Of Calling Your Co-Founder

Instead of building a CMS or database first, Pontus shipped a hard-coded JSON file with rules scraped from online. Within 30 minutes the site was live on Vercel and clickable; total build to launch was 3 hours flat on Next.js with a reused design system. The data layer can always upgrade — the bar to validate is just 'is it live?'.

Launching
I posted the cursor directory on X at the initial release. I mean we're building public on X, we share everything we do all the way to the code right, we are fully open source. So I think we had like 1 million impressions on one of the posts and of course we had several posts so we have a developer following from the beginning.

Stack Months Of Build-In-Public On X So One Launch Post Hits A Million Impressions

The launch wasn't a cold drop, it was the payoff of an existing build-in-public habit on X with an open-source developer audience. One post hit 1M impressions, then Hacker News front page, then YouTuber coverage compounded on top. The pre-built distribution is what made a 3-hour weekend project flip to $35K/month inside weeks.

Content
It's open source all the rules are merged on in GitHub so the maintenance cost is like 3 hours per month and then we continue building our main startup.

Let The Community Submit Content Via GitHub Pull Requests So The Directory Grows Itself

Cursor Directory's content layer is a public GitHub repo where users PR new rules — the founders just review merges. That turns content production into a community input pipeline, keeps maintenance at 3 hours/month, and gives every contributor a reason to share the site with their audience.

Pricing
companies can share their job ads so it's a paid ad that they pay to get listed some months after MCPS came along and that were also a thing... these are also featured so you can pay to get them featured on the site

Monetize Directories With Paid Featured Listings For Tools, Jobs, And New Integrations

Cursor Directory monetizes through paid placements — companies pay to list job ads, and pay to feature their MCP integrations on the site. Layering a second ad surface (MCPs) on top of the original (jobs/rules) is what unlocked the $35K/month run rate at 99.8% gross margin.

Onboarding
we also made it possible to generate your own cursor rules tailored to your website in JavaScript you have this package JSON which defines all your dependencies you could just upload that and then you get your defined cursor rules back... developers wanted to tailor their cursor experience but they didn't know how to start this was a good way for developers to get the grasp of it all

Solve First-Visit Cold-Start By Generating Personalised Output From Whatever The User Uploads

They added a personalization tool where developers upload their package.json and get back custom cursor rules matched to their dependencies. This solved the 'don't know where to start' cold-start problem for first-time visitors and gave them immediate, personal value before asking for anything — turning a directory into a tool on the first click.

Product
Directories still are incredibly successful in 2025 especially if you build something for the right audience... the directory solves a timeless problem like you need to find something that you're interested in and in this new AI world where we live in like cursor, people are starting to look through how to do things... if you can catch that wave you're really going to find a lot of visitors.

Directories Aren’t Dead — They Win By Catching A New Platform Wave First

The directory model still works when it rides a new platform wave (Cursor, then MCPs) before anyone else has aggregated the content. Pontus added a second adoption spike by making MCPs searchable on the site right when that ecosystem emerged — owning the canonical reference for a hot new category compounds traffic over time.

Bootstrapping
We chose Next.js fully TypeScript shadcn components, resend for email, OpenPanel analytics, and just connected with GitHub and shared it on Vercel... payments we use Polar... every tool we choose to share have the same philosophy: no servers, no friction, just focus on building and shipping fast... all in all 525 bucks a month with a 99.8 gross margin.

Run $525 Of Managed-Service Tools Against $35K Revenue For A 99.8 Percent Gross Margin

The entire stack is managed services with no servers to babysit — Next.js on Vercel, Resend, Polar, shadcn. Total infrastructure runs $525/month against $35K revenue, and the open-source GitHub-merge model means contributors do most of the content work, keeping founder maintenance to 3 hours/month while two people run the directory on the side.