vasusen

It was for use by our paid customers in CI/CD. We did not want large companies to use it as is without a contract from us.

We moved our SDK off public npm onto a private registry we host.

Our customers mainly use Donobu's SDK to run their end-to-end tests in CI/CD, with AI self-healing and triage. It was unlicensed but on public npm on purpose, so `npm i` just worked.

NPM registry has had a rough year. When we looked at alternatives, we realized self-hosting packages is not that hard anymore. Our customers continue using `npm i`. They just add a .npmrc pointing at our registry; reusing their existing Donobu API key for auth was a nice bonus.

.npmrc was the only thing needed:

  @donobu:registry=https://api.donobu.com/npm/
  //api.donobu.com/npm/:_authToken=${DONOBU_API_KEY}

Under non-urgent cases this sometimes takes 3-4 months in the US every time I experience the need to "ask an MD"

I am really excited for this. Once going home with my family via Uber in SFO we realized on the freeway that our driver was high and driving at 80-85 mph.

It was a really scary experience and I couldn’t do much about it in the moment.

They need to make it eligible for Class Action lawsuits to be filed if these are ignored. I wrote a script to routinely test opt out on websites and was stunned to see almost 50% had it implemented incorrectly. This includes high-flying tech companies that went public recently.

Under California’s CCPA / CPRA, most enforcement power lies with the California Privacy Protection Agency (CPPA) and the California Attorney General, not private individuals. This limits the actual downside to a company vs. an unbounded downside of class-action lawsuit threat.

Like many things where what constitutes base data changes and hence graphs change pretty radically, this can be cleanly explained by diagnostic drift. Here's a detailed breakdown: https://www.cremieux.xyz/p/how-to-end-the-autism-epidemic

Highly recommend this Coursera course from University of Michigan, "Influencing People": https://www.coursera.org/learn/influencing-people

Super cool! I wonder if it would bother pets

The reason isn't technical. This isn't implemented because the entire card-processing ecosystem is hooked on the chargeback fees (min $15 to $100). It starts becoming a lucrative revenue stream for Visa/Mastercard/Stripe/Adyen/WorldPay/Fiserv and the entire ecosystem.

Merchant's end up getting the short end of the stick in most cases.

It is quite popular in testing circles to write e2e tests that are easier to maintain. However, in practice I have found it to be quite useless due to the time it takes to write good page objects. QA teams usually rely on a complete POM before writing tests on it. I used to joke that by the time my team was done shipping a page object model, our product team would have changed the entire product again.

I used to be in-charge of homepage getting over 1.5M views a day. I would really be curious how this converts. I am assuming Posthog has a lot of metrics.

If I were to bet, while this is fun, it will be a disaster for conversions once the launch hype goes away.

Yes an electron app helps tremendously, especially for managing lifecycle of tabs independently. We use that for creating our AI browser automations at Donobu (https://donobu.com). However, we do have the luxury of just focusing on a narrow AI QA use case vs. Browser-Use and others who need to support broad usecases in potentially adversarial environments.

2011 were definitely not the dark ages!! I used to use Selenium for everything back in the day. I was able to scrape all of Wikipedia in 2011 entirely on my laptop and pipe it to Stanford NLTK to create a very cool adjective recommender for nouns.

Chewing plant twigs to clean teeth is an ancient way of tooth cleaning in many cultures[1]. I wonder if the Lignin or Suberin in plants acts the same was as Keratin in this study.

1. https://en.wikipedia.org/wiki/Teeth-cleaning_twig

I love this website.

It doesn't have to be all or nothing. Some AI tools can be genuinely helpful. I ran a browser automation QA bot that I am building on this website and it found the following link is broken:

"Every Layout - loads of excellent layout primitives, and not a breakpoint in sight."

In this case, the AI is taking action on my local browser at my instance. I don't think we have a great category for this type of user-agent

My wife decided to not travel to Japan due to an impending warning from a manga for July 2025. I have been making fun of her all month only to get this tsunami warning now!

https://en.wikipedia.org/wiki/July_2025_Japan_megaquake_prop....

The 2021 reprint capitalizing off this revived popularity warned of a "real disaster" in July 2025, causing a minor case of mass hysteria in 2025 when summer trips to Japan from East Asia decreased markedly and several airlines even cancelled flights.

I grew up in India and now live in the US. My mom recently got some ground turmeric from our own farm when she visited us. I am was stunned by how much more duller, brownish-yellow it was compared to the turmeric I buy in Indian stores in the US. Those are usually really bright yellows.

Now, I am really scared that even stuff sold in California is probably lead paint tainted turmeric.

I think the Playwright MCP is a really good example of the overall problem that the author brings up.

However, I couldn’t really understand if he’s saying that the Playwright MCP is good to use for your own app or whether he means for your own app just tell the LLM directly to export Playwright code.

I am not from the deaf community, but my son has severe hearing loss. I really look forward to a world where condition like his simply becomes fixable like bad teeth and he doesn't have to miss out on so many things.

thank you, I appreciate that!

I love the idea of a trainable robot arm as a learning device at that price point.

However, seeing the chess demo instantly makes me think of that horrible tragedy with the robotic arm breaking a kid's finger. How strong is this to be used around kids?

I do this too by pasting only the relevant context files into O3 or Claude 4. We have an internal tool that just lets us select folders/files and spit out one giant markdown.

Grok's unhinged mode is the closest I have seen a bot that leverages profanity. I find it quite entertaining to use occasionally like watching a South Park episode.

Thank you so much for creating this folks! A browser navigation agent is key part of our AI QA setup at Donobu (https://donobu.com/). We found the WebVoyager benchmarks severely lacking for complex e2e test cases like logged-in dashboards, onboarding forms, etc.

While the extraction/2fa flows aren't super relevant to us, this saves us time from building our own set of benchmarks. Really appreciate it and hope we can contribute to make this a really large set.

Thanks for building it! It can really help help my 10 yo who is just learning coding. How does it handle potentially infinite loops like below? Currently I get a parsing error after it takes a while.

```python

  while exit != "yes":
    print("\*")
    exit = input("Exit?: ")

(disclosure: I am no longer at Coursera)

When Duolingo added that viral post on Growth hacking, it caused quite a stir about the push-notifications and gamification tactics they use. Ultimately, we decided it wasn't worth it for Coursera to veer into edu-tainment.

However, it is interesting to watch how much gamification works in adding and retaining users. In 2023, Duolingo's marketcap was 5x of Coursera. Now at similar revenue it is 20x of Coursera.

As a user, I think Duolingo is over-gamified (stopped using it) but Coursera is severely on the other spectrum where it comes off as too bland/boring to keep up the motivation. I am sure there's a happy medium to be found between reminding users to engage in something hard while doing right by learners.

What worries me most about MCP isn't that the protocol is so poorly created, but that fixing/improving it is only at the mercy of internal teams at Anthropic and OpenAI. It doesn't seem like the folks who are coming up with the protocol are actual engineers trying to implement it as well.

Vaguely seems like a Visa-Mastercard duopoly.

Sorry instead of "googling" this plant up, I used ChatGPT to tell me candidates on what this plant might be.

Congratulations on the launch!

How do you get the medical professionals to see a "diff" of the changes to approve? Is there versioning in the EHR tools?

I believe (more likely, chatgpt believes) this plant is Warabi. I didn't know it was processed with charcoal purification!

The process sounds similar to how I use something acidic to process the toxins from Taro leaves before eating them.