DEV Community: Mike Alex

Is GitHub Trending Just AI Agents Now? 🤖

Mike Alex — Wed, 28 Oct 2026 10:47:36 +0000

I opened GitHub Trending this morning, like I do most days.

And I noticed something.

Every single repo on the list was about AI.

Not most of them. All of them.

Agent frameworks. Agent skills. Agent workspaces. An AI tutor. Even an IDE built around vibe coding.

Here's what was trending today

Repo	What it does	Stars today
Nutlope/hallmark	Anti-AI-slop design skill for coding agents	2,830
HKUDS/DeepTutor	Personalized AI tutoring	1,598
makecindy/cindy	Open-source AI agent that works out of the box	1,465
pbakaus/impeccable	Design vocabulary for your coding agent	1,449
elayadesign/ai-design-skills	Design skills for AI agents	1,395
JetBrains/thinkrail	A lightweight IDE for vibe coding	1,204
affaan-m/ECC	Performance system for agent harnesses	1,165
nightly-labs/openbot	Local-first workspace for AI teammates	777
obra/superpowers	Agentic skills framework	617

(From GitHub Trending on Oct 28, 2026. These numbers change fast.)

11 out of 11 repos were about AI.

That's not a trend anymore. That's the whole page.

The part that made me laugh

Look at the top of that list again.

The #1 repo today is an anti-AI-slop design skill. Two more in the top five exist to stop coding agents from shipping generic, ugly UIs.

So we're now building AI tools… to fix what other AI tools build.

Honestly? I think that's a good sign. It means people are moving past "look, the AI made a website" and starting to care whether it's actually good.

Is an all-AI trending page a bad thing?

I'm not sure.

On one hand, it's exciting. A huge amount of open-source energy is going into one idea, and things are moving incredibly fast.

On the other hand, I keep wondering:

How many of these will still be maintained in six months?

A lot of repos get thousands of stars in a week. Fewer get issues closed, docs written, and versions released.

Stars show interest.
Commits show commitment.

What I've started doing

Before I star or try a trending repo now, I check three things:

Last commit date. Is anyone still working on it?
Open issues vs closed issues. Are problems actually getting fixed?
The README. Does it explain what it solves, or just how impressive it is?

It takes about 30 seconds. And it has saved me from a lot of abandoned projects.

Meanwhile, what's getting missed?

This is the part I'm most curious about.

When one topic takes over the entire page, great projects in other areas get buried. Databases, dev tools, accessibility, games, plain useful libraries.

So I'd love to hear from you:

What's a non-AI repo you found recently that deserves more stars?
Do you trust GitHub Trending to show what's actually good?
Have you ever starred something trending and regretted it?

Drop your hidden gems in the comments.

I'll try every single one and share the best ones in a follow-up post 👀

What Should Beginners Learn First in 2026: Fundamentals or AI Tools? 🤔

Mike Alex — Tue, 27 Oct 2026 07:06:55 +0000

A beginner asked me something last week that I haven't stopped thinking about.

"Should I learn JavaScript properly first, or just start building with AI?"

I didn't have a clean answer.

Because both sides make sense.

The case for fundamentals first

If you don't understand what the code does, you can't tell when AI gets it wrong.

And AI gets things wrong in quiet ways. The code runs. The tests pass. But the logic is off, or there's a security hole, or it breaks the moment real users show up.

Someone has to notice that.

Fundamentals are what let you notice.

The case for AI tools first

On the other hand, motivation matters.

A lot of beginners quit during the "build a to-do list for the fifth time" phase. They never get to the part where coding feels exciting.

With AI, a beginner can build something real in a weekend. Something they actually care about.

And that excitement might be what keeps them learning long enough to get to the fundamentals.

Where I've landed (for now)

I'm a product manager who builds with AI every day. I'm not a classically trained developer.

Here's my honest experience:

AI made me fast.
Understanding the basics made me useful.

The times I got stuck weren't when I couldn't write code. They were when I couldn't read it. When something broke and I didn't know where to look.

So maybe the question isn't "fundamentals or AI."

Maybe it's: build with AI, but never accept code you can't explain.

Ask the AI why. Break things on purpose. Rewrite small parts by hand.

Use AI as a teacher, not just a shortcut.

But I want to hear from you

This is one of those questions where experience really changes the answer.

If you're a beginner: Are you learning with AI or without it? How's it going?
If you're experienced: If you were starting from zero in 2026, what would you do first?
If you hire developers: Does it matter to you how someone learned?

I don't think there's one right answer.

But I'm pretty sure the comments will be better than the post 😄

If AI Can Build an App in an Hour, What Makes a Developer Valuable Now? 🤔

Mike Alex — Sat, 10 Oct 2026 13:01:07 +0000

If AI Can Build an App in an Hour, What Makes a Developer Valuable Now? 🤔

A thought I've been having lately.

With AI tools, you can build a website, create an app, or turn an idea into a working prototype in just a few hours.

Things that once took days can now take much less time.

So here's something I'm curious about.

If almost anyone can build an app with AI, what actually makes a developer valuable now?

Is it still how well you write code?

Or is it something else?

Maybe it's knowing what to build in the first place. Understanding the problem, making the right technical decisions, debugging when things go wrong, or building something people actually want to use.

Because getting an app to work is one thing.

Making it reliable, useful, maintainable, and worth using is another.

I'm not saying coding skills don't matter anymore. But I do wonder how the definition of a good developer is changing as AI gets better.

I'd love to hear your thoughts:

What skill makes a developer stand out in the age of AI?
Would you hire someone who can build quickly with AI but doesn't understand the underlying code?
What are you focusing on learning to stay valuable as a developer?

I don't think there's one right answer, and that's what makes this interesting.

I Tested 16 Job Sources With a Scraping API, here's What Worked!

Mike Alex — Fri, 09 Oct 2026 12:14:05 +0000

I Tested 16 Job Sources With a Scraping API. Here's What Worked

Job hunting can get messy quickly.

You find a role on LinkedIn, another on a startup's career page, and a few more on remote job boards. Soon, you have dozens of tabs open and no single place to track everything 😅

I started thinking about building a small tool that collects job listings from different sources and organizes them in one place.

As a product manager working with Spicrawl, I wanted to explore how its web scraping API could help.

In our October 2026 tests, 15 out of 16 job sources returned usable listings. Here's what we learned.

1. Not every website needs a browser

Some job sources provide data through public APIs or RSS feeds. Others rely on JavaScript to display listings.

Here's a snapshot of our results:

Source	What worked	Credits
Remote OK API	100 job items	1
Remotive API	Structured job data	1
LinkedIn single posting	Full job description	1
LinkedIn job search	Rendered listings	3
Y Combinator jobs	Rendered listings	3
Glassdoor jobs	Rendered listings	3
Indeed search	Listings after scrolling	8

These are results from our tests, not guaranteed results for every request.

The takeaway? Start with the simplest request and only add browser rendering when you need it.

2. Collect job listings with Spicrawl

You'll need a Spicrawl API key, which you can get through Spicrawl.

For a JavaScript-rendered job page, you can use:

curl https://api.spicrawl.com/v1/scrape \
  -H "Authorization: Bearer $SPICRAWL_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "url": "https://www.ycombinator.com/jobs",
    "js_render": true,
    "cache": false
  }'

This loads the page in a browser so its JavaScript can build the listings.

In our October tests, this returned around 28 KB of content for three credits.

For a source that already provides JSON, you can use a plain fetch instead. For example, Remote OK returned 100 items for one credit.

3. What about pages that load more jobs while scrolling?

This was one of the more interesting results from our tests.

Indeed returned only a tiny, 60-byte page when we tried JavaScript rendering alone. The request technically succeeded, but the job listings weren't there.

We added scrolling actions:

{
  "url": "https://www.indeed.com/jobs?q=data+engineer&l=Remote",
  "js_render": true,
  "wait": 2000,
  "actions": [
    {"scroll": {"to_bottom": true}},
    {"scroll": {"to_bottom": true}}
  ]
}

The response grew to around 68 KB of listings. This request cost eight credits because browser actions use the full-browser rendering tier.

A successful HTTP response doesn't always mean you received useful data. Check the returned content and target status before saving results.

4. Turn scraped pages into a job tracker

Once you've collected the page content, you can extract fields such as:

Job title and company
Location and employment type
Salary, when available
Required skills
Application URL

Spicrawl's autoparse: true option can return structured data embedded in a page, including schema.org JobPosting data when available.

You can store the results in a database or spreadsheet and build features such as job filters, application tracking, and AI-powered comparisons against your skills.

Just remember that not every page contains every field, and some job listings may be outdated.

5. What we learned

A few lessons stood out:

Use APIs and feeds when available. They're often cheaper and easier to process.
Render JavaScript when needed. Some listing pages don't contain their data in the initial HTML.
Check the actual response. Empty pages can still return HTTP 200.
Respect website rules. Check terms and robots.txt, follow rate limits, and collect only the data you need.

What would you build?

A job collector is just one use case. The same idea could help organize internships, scholarships, research opportunities, or listings for a job aggregator.

If you were building a job-hunting assistant, what would you want it to do first?

Find relevant jobs, compare opportunities, match jobs to your skills, or track applications?

I'm exploring these use cases while working on Spicrawl, and I'd genuinely love to hear your ideas.

Want to explore more job boards? Check out the Spicrawl job-postings scraping guide for tested API requests, credit costs, JavaScript rendering, and tips for scraping different job sources.

Explore Spicrawl currently free during beta, with no credit card required.