8 generator tools compared after 3 months of daily use
The eight generator tools I actually keep open are a flexbox playground, plus generators for CSS gradients, box shadows, color palettes, favicons, QR codes, passwords and SVG paths. Each one earns its tab for quick one-off work. I swapped the favicon generator for a script on one project that needed repeatable builds, and that script is below.
Quick disclosure: the generator tools I link to below are ones I built. Over about two years I bookmarked 11 different sites for gradients and favicons, and nearly all of them wanted an email before they'd hand over the output, or quietly uploaded my logo to a server. Mine run client-side in your browser. They're free, with no signup and no upload. If you have a better one, tell me.
Why generators ended up pinned in my browser
In the first week of August 2026 I rebuilt the landing page for a small invoicing app I run on the side. Nothing fancy. A hero section and a pricing grid. I'd budgeted one afternoon.
It took four days. When I went back through my browser history afterward, I worked out that I'd spent around 47 minutes, spread across those days, nudging box-shadow values one pixel at a time and hitting reload. Forty-seven minutes on shadows. On a page with five cards.
Generator tools are boring, and that's probably why nobody writes about them. You open one, drag a slider, copy six lines of CSS and close the tab. The time they save is real, though, and it adds up when you do visual work in short bursts between backend tickets (which is how most of us full-stack-by-necessity people end up working).
So after that rebuild I started paying attention to which generators I actually opened. I ignored the ones I'd bookmarked with good intentions and only counted the ones I used. Three months later, eight survived.
Some of them surprised me. I honestly thought I'd never touch a QR code generator again, and then in September a client asked for printed flyers for a trade show and needed the code by 4pm. I was wrong about that one.
The tools I kept using
Five of these get real detail below. The rest get a line in the comparison table, because there isn't much to say about them beyond "it works and it's fast."
Flexbox playground
I've written flexbox for nine years and I still forget which property aligns along which axis the moment flex-direction becomes column. A playground fixes that faster than any docs page, because you watch the boxes move as you flip the direction.
- What I use it for: reproducing a broken layout in isolation, then copying the working properties back into the real stylesheet.
- The catch: it's flexbox only. For real two-dimensional layouts I go to Grid and read MDN like everyone else.
CSS gradient generator
Gradients are the thing I'm worst at by eye. I'll write linear-gradient(135deg, #6366f1, #a855f7), decide it looks muddy in the middle, and then spend ten minutes guessing at a third color stop. A generator lets me drag that stop to 38% and see the result immediately, which is the whole point.
My one rule: if two colors turn grey where they meet (purple into orange is the classic offender), add a midpoint stop with a more saturated color between them. And if any generator anywhere hands you -webkit- prefixed gradients, delete them. Every current browser handles the unprefixed syntax.
Box shadow generator
This is the one that started the whole exercise, so it gets the most opinion.
Single shadows look cheap. Layered shadows, two or more stacked with increasing blur and decreasing opacity, look like something a designer made. Something like box-shadow: 0 1px 2px rgba(0,0,0,0.06), 0 4px 12px rgba(0,0,0,0.08); reads as "soft card" instead of "2009 drop shadow." Writing those by hand is miserable because every value interacts with the others.
- Build the shadow in the generator, then paste it into a CSS custom property.
- Keep the token count tiny. My invoicing app has exactly two,
--shadow-cardand--shadow-popover, and I haven't needed a third. - Check it on a phone. Heavy layered shadows that look lovely on a 27-inch monitor can look smeared on a small screen.
Color palette generator
I'm going to be honest: I don't fully trust any palette generator, including mine. They're great at producing five colors that look nice next to each other as swatches. They're less helpful at telling you whether white text on swatch number three is readable.
So I treat it as a starting point. A client says "make it feel friendly," I generate a few options, pick one, and then run every text/background pair through a contrast checker before anything ships. WCAG AA asks for 4.5:1 on normal body text, and a surprising number of pretty palettes miss it.
Password generator
This one's about trust more than convenience. I use it for test accounts, seed data, staging database passwords and the occasional API secret I need right now and will rotate later.
- The check I'd run on any browser password generator, mine included: open the network tab in devtools, click generate, and confirm nothing gets sent anywhere. If you see a request fire, close the tab.
- The randomness should come from
crypto.getRandomValues.Math.randomis predictable enough that it has no business anywhere near a password.
Side-by-side comparison
All eight live together on the aidevhub generators page, which is the one link I'd actually bookmark. Here's how they stack up after three months, including the honest weak spot for each.
| Tool | Best for | Pricing | The one dealbreaker |
|---|---|---|---|
| Flexbox playground | Debugging alignment and axis confusion | Free | Flexbox only, no Grid |
| CSS gradient generator | Fixing muddy color transitions | Free | Won't stop you from picking ugly colors |
| Box shadow generator | Layered, soft card shadows | Free | Easy to overdo, so test on mobile |
| Color palette generator | Starting points for a new theme | Free | Pretty palettes can still fail contrast |
| Password generator | Test accounts and staging secrets | Free | It generates but doesn't store; you still need a manager |
| QR code generator | Flyers, slides, quick links | Free | Static codes only, so a URL typo means reprinting |
| SVG path editor | Tweaking small icons by hand | Free | Painful past a couple hundred points |
| Favicon generator | One-off sites and prototypes | Free | Manual, so it isn't repeatable in CI |
That last row is where I'll be the most self-critical.
The one I ditched and why
In mid-October I was on a client project where the logo changed constantly during design review. I regenerated favicons 6 times in 9 days, and on the fourth round I shipped a stale 32px icon to staging because I forgot to replace one file. The client noticed before I did. Honestly, that annoyed me for a full day, mostly because it was my own tool and my own sloppiness.
So for that project I ditched the browser tool and put favicon generation in the build. It's about 20 lines of Python with Pillow:
# make_favicons.py
# Usage: python make_favicons.py logo.png out/
# Assumes a square source image, ideally 512px or larger.
import sys
from pathlib import Path
from PIL import Image
src_path = Path(sys.argv[1])
out_dir = Path(sys.argv[2])
out_dir.mkdir(parents=True, exist_ok=True)
logo = Image.open(src_path).convert("RGBA")
# Multi-size .ico for browser tabs and older clients
logo.save(out_dir / "favicon.ico", sizes=[(16, 16), (32, 32), (48, 48)])
# PNGs for apple-touch-icon (180) and the web manifest (192, 512)
for size in (180, 192, 512):
icon = logo.resize((size, size), Image.Resampling.LANCZOS)
icon.save(out_dir / f"icon-{size}.png")
print(f"Wrote favicons to {out_dir.resolve()}")
Run pip install pillow first, then call it from whatever build step you already have. Now the logo file is the single source of truth, and a new revision means the icons regenerate on the next deploy. No human in the loop to forget a file.
I don't know whether the 48px entry in the .ico still matters for anyone in late 2026. I keep it because removing it feels like tempting fate, and it costs a few kilobytes.
Would I tell you to skip the browser favicon generator? No. For a prototype, a hackathon project or a personal site where the logo won't change, opening a tab is faster than wiring up a script. The moment an asset changes more than twice, though, automate it. That's the lesson I keep relearning with generators in general: they're perfect for decisions you make once, and the wrong tool for decisions you'll make every week.
FAQ
Q: Are browser-based password generators actually safe to use?
A: They can be. Check that the tool makes no network requests when you generate (devtools network tab, ten seconds) and that it uses crypto.getRandomValues. If you can't verify either, use your password manager's built-in generator instead.
Q: Why not just ask an AI model for the CSS?
A: I do sometimes, for gradients especially. For shadows I find a slider loop faster than re-prompting "a little softer, no, less than that" five times. Visual tweaking wants visual feedback.
Q: Do QR codes from a free generator expire?
A: Static QR codes don't expire, because the data is encoded directly in the pattern. The ones that "expire" come from dynamic services that route scans through their own redirect URL, and they can switch that redirect off whenever they like.
Q: Which one should I try first?
A: The box shadow generator. It's the one where the gap between hand-written and generated output is the most obvious, and you'll see it on the first card you restyle.
Written with AI assistance and human review. Try the tool at aidevhub.io/tools/generators.
Top comments (0)