Vibe coding is not a silver bullet.
The value you get from coding with AI depends a lot on the developer and the problem you're trying to solve.
@gethackteam.bsky.social
AI, Web Dev, Developer Experience & DevRel @ibmdata β‘ π₯ Tech Author, Speaker, Entrepreneur π Sharing opinions (βοΈ hello@hackteam.io
Vibe coding is not a silver bullet.
The value you get from coding with AI depends a lot on the developer and the problem you're trying to solve.
The 4 levels of vibe coding
(this sounds as a joke, but itβs actually how people build now)
SF does not just have tech eventsβ¦
First marathon β
Thank you, ChatGPT
25.07.2025 12:34 β π 3 π 0 π¬ 0 π 0I don't trust developers that use test coverage percentages as quality score for their project
23.07.2025 18:38 β π 4 π 0 π¬ 1 π 0Junior Developers are doomed...
Says the guy who doesn't know how to code and did not try AI to vibe code himself
According to experts AI will ruin the world.
What did the same experts say about Y2K?
Doomsday.
Developers love to overengineer.
Thatβs why I like vibe coding. Just build something simple and get it in front of people fast.
And worry about scaling when you actually do reach scale.
Stars and downloads signal hype or interest but they donβt prove real-world adoption or value.
If you care about adoption, you need proper usage metrics. Not vanity numbers.
People star repositories for all sorts of reasons:
- To stay updated on changes
- To bookmark it for later
- To show support
And downloads are inflated too: automated trackers, bots, and CI/CD systems all inflate the counts.
That means the JS version has a much higher stars-to-download ratio β about 3Γ more stars per 1,000 downloads than Python.
What does this tell us? Not much.
Thereβs no causality between starring a repository and actually using the library...
The Python version has been around longer and has ~111k stars, while the JS version has βonlyβ ~15k stars.
But if you look at the ratio between downloads and stars, the story flips:
- Python: ~0.0058 stars per 1,000 downloads
- JS: ~0.0163 stars per 1,000 downloads
What does this tell us?
GitHub stars, npm and PyPI downloads look impressive on paper, but they donβt really measure real usage.
Letβs take LangChain as an example β which has both a Python and a JS version.
GitHub stars donβt tell you how much a library is actually used. Not even close.
14.07.2025 18:38 β π 2 π 0 π¬ 1 π 0Thinking is becoming a skill again
14.07.2025 17:20 β π 5 π 0 π¬ 0 π 0Every PM should vibe code
14.07.2025 15:05 β π 2 π 0 π¬ 1 π 0Race for the best models, who will win?
10.07.2025 18:38 β π 2 π 0 π¬ 1 π 0Check out the full video here www.youtube.com/watch?v=TMmi...
08.07.2025 12:11 β π 1 π 0 π¬ 0 π 0MCP Inspector is like Postman, but for MCP servers instead of APIs
08.07.2025 12:11 β π 3 π 0 π¬ 1 π 0If you could automate half your job, would you waste the time or use it to grow?
The way you spend that freed-up time will define your future.
Awesome MRR for a βboringβ company
30.06.2025 15:16 β π 3 π 0 π¬ 0 π 0My number 1 employee
28.06.2025 16:58 β π 3 π 0 π¬ 0 π 0AI is not just for coding
28.06.2025 09:52 β π 4 π 0 π¬ 1 π 0Everyoneβs chasing the next agent framework.
Few are asking what problems are actually worth solving.
The future isnβt no-code.
Itβs high-context code.
Learning to code was just step one.
Now you have to learn how to explain code to a machine that writes it for you.
Coding used to mean "write every line."
Now it means "design the outcome."
You donβt need to remember how to reverse a binary tree anymore.
But you do need to know when one makes sense.
Knowing what to build and why just became more valuable than knowing how to build it.
14.06.2025 18:38 β π 2 π 0 π¬ 0 π 0Check out the video and tutorial
- www.youtube.com/watch?v=TMmi...
- hackteam.io/blog/build-t...