Slopdar
Is it built, or is it slop?
A few days after launching Slopdar, I kept seeing the same comment:
"My website was built entirely with AI, but Slopdar gave it a low score."
At first, I thought people had found a bug.
But they hadn't.
The confusion came from assuming Slopdar was answering the question:
"Was AI used to build this?"
It isn't.
The question Slopdar tries to answer is:
"Did AI leave fingerprints behind?"
Those are very different things.
A founder can build an entire product with Lovable, Cursor, Claude, GPT, or Bolt, then spend hours rewriting copy, improving the UX, replacing generic components, and polishing every detail. The result may not feel AI-generated at all. I'd expect that site to receive a relatively low Slop Score.
Another founder can use the exact same tools, accept the first output, and ship it with generic layouts, predictable copy, default components, and obvious AI patterns still intact. In that case, the fingerprints are everywhere, and the score should be much higher.
That's why Slopdar doesn't just return a number.
Every scan includes the reasoning behind the score. I don't want people to blindly trust 72/100. I want them to see what the tool actually found and decide whether they agree with it. If someone disagrees, that's useful feedback rather than a failure.
Ironically, some of the best feedback has come from people trying to break it.
Every false positive, every false negative, and every "this should have scored higher" becomes another test case. That's how the scoring improves.
I don't think AI is the problem. I use AI every day, and I'd happily use Lovable or Cursor to build my next project.
What I'm interested in is the difference between using AI well and shipping the first thing AI gives you.
That's what Slopdar is trying to measure.
I'm curious how others think about this.
If you were building a tool like this, would you focus on detecting AI usage, or AI fingerprints?
I kept seeing AI text detectors.
But I couldn't find a tool that analyzes an entire website.
So I built Slopdar.
Paste any URL and it:
🎯 Gives a Slop Score (0-100)
🧾 Shows the receipts behind the score
😂 Roasts the website for fun
It's completely free and doesn't require signup.
It started as a weekend experiment.
A few days later:
🚀 Launched on Product Hunt
🌐 Hundreds of websites scanned
💬 Lots of feedback (and bug reports 😅)
One of my favorite moments was when someone pointed out a false positive.
I fixed it the same day, and they came back offering to help improve the heuristics.
That's exactly why I love building in public.
If you want to roast your own website (or your friend's 😄), I'd love your feedback.
-
1
This is fun and interesting, the reasoning is cool and genuinely useful. I'm not sure about the scoring though - is it just for fun or do you want it to be good?
-
1
Thanks! I definitely want the score to be useful, not just a fun number. It's still early, so I'm tweaking it as more people try it and share feedback. That's also why I show the reasoning. I'd rather people understand why a site got a score than blindly trust the number.
-
2
So I tried my two sites and the I got a lower score on the vibe coded one than on my old webflow one. Anyone once again cool work!
-
-
-
1
Pretty cool work! Think this can be very useful for basic fraud detection. Like - let's see who's not professional... Just one thing to note. I slapped in my fully lovable made prototype (check my profile, for Rogue Headlines, you'll see it immediately). It gave it a 56 score even though I'd have expected it to be considered maximum sloppiness (I don't know how to code and didn't code anything...) haha. Other than that, think this thing could become quite useful
-
1
That's actually by design. Slopdar isn't checking whether AI was used to build a site. It's looking for AI fingerprints. A site can be built entirely with Lovable and still score low if those fingerprints have been removed. I'd love to see your prototype if you're happy to share it. It would make a great test case.
-
-
1
I like that you're showing the reasoning behind the score instead of just assigning one.
The interesting challenge is making sure people use the score as a starting point for investigation rather than treating it as an objective measure of quality. That distinction will probably matter more as the product grows.
-
1
I completely agree. The score is just a summary. The reasoning is what I want people to focus on. That's why every scan includes the receipts instead of asking people to trust a black box.
-
2
Interesting.
Your reply made me think less about the score itself and more about what the product quietly commits itself to once the reasoning becomes the thing users are really expected to trust.
I don't think I can explain that line of reasoning properly in a thread without oversimplifying it.
If you're interested, what's the best email to reach you on?
-
-
Comment