Have been using Claude successfully for two days to fix the date issue I had on my On This Day page. It was a long process that involved forking my own script, changing the JS code and Python (highly tailored to my case, because I had some elements I added in my Hugo theme), hosting my own instance on Render, and then going back and forth (which took the most time) to look for bugs with Safari’s Inspector, using the console. All of this was done with the help of Claude.

The next step is to go back and write a thorough post covering what was done and why. This stuff keeps blowing me away.

Manton about AI usage in Micro.blog:

Earlier this year I blogged a strategy for how I want to use AI thoughtfully in Micro.blog. It has been a good guide for me, like user-centered guardrails. There is still so much we can build that fits within that strategy, hopefully avoiding the worst “put AI in everything” fixation from bigger companies that users are rejecting.

Micro.blog has a master switch for AI. If the user turns it off, there’s no AI usage. Period. None of the AI features would work:

Auto-generated description: A notice explains that Micro.blog uses artificial intelligence and OpenAI for various features but ensures data privacy and non-use for training.

While it’s true that users of Micro.blog would lose some of the benefits of the service if they turn it off (and they’re paying for the service just the same), it is right to let the users choose if they want to use AI, without forcing it everywhere possible, like Microsoft has done with Copilot.

I find that Micro.blog has followed this mentality, especially looking into what is being done around Inkwell (Inkwell has a feature where AI can summarize several RSS feeds in a helpful “here’s what happened on JTR’s blog in the last week” sort of way).

This post from Manton didn’t come out of nowhere; he was listening to Nilay Patel’s opinion piece about “Software Brain,” which is worth listening to. It’s one of the reasons I enjoy following Manton’s feed - it’s not just Micro.blog, but there are also interesting opinions about AI and using it.

On This Day page dates are broken

When you fix one thing, you break another. ISO dates are nice, but now my “On This Day” page doesn’t understand what these mean. American dates have and will always be an “old man yells at cloud” issue for me.

Doctorow reflects on how, in a way, his life’s work can be summed with one angry poop emojy:

Pluralistic: A free, open visual identity for enshittification (24 Apr 2026) pluralistic.net

Needless to say, I got myself the sticker.

So far, I’m enjoying Inkwell. The main advantages over my FreshRSS setup is mainly the speed and the cleaner UI.

Since I discover most of my RSS feed via blogs and reading posts online, I already have the browser open, and adding a new feed to Inkwell is faster then doing so on FreshRSS because my Synology needs to wake up if it’s hibernating, then it needs to log me in - and sometimes the container FreshRSS is on is not available for some reason and I need to restart it or wait patiently. Not an issue with Inkwell.

There are a few issues that need to be polished out, especially on the web and the Mac version, but I’m already leaning into the philosophy behind the app: read what’s new today, and don’t worry about what I missed. It’s easy enough to visit the Inkwell blog and read past posts if needed.

Things I want to do with my website (part 1)

I want to do more with my website. I play 2 plus hours of Linux games daily, but blog about them once a month, maybe. Why? Because one game deserves 5 posts, not a paragraph. I kind of have a solution, just not the time. A familiar pattern, I know

About writing other bloggers Email

Writing emails to each other as bloggers is good.

Testing Inkwell by Micro.blog

Running FreshRSS on Docker synced to Elfeed has served me well, but since I actually read most feeds on my Android, Inkwell’s beta is worth testing. Micro.blog expanding into RSS reading makes sense, and it looks pretty good so far.

Archive by month

So I guess I am on a roll?

Added a small square as a decorative element (a little square) for the dates in the posts. That’s the small change.

I also figured out why Manton’s plugin, for archive by month, didn’t work for me. It clicked when I realized what went wrong last time.

The plugin is meant to replace the default layouts/list.archivehtml.html, not the one I have, which is slightly modified by TinyTheme. So what I just went and snatched the code from the above and pasted it inside layouts/list.archivehtml.html, write after the condition to activate the microhook partials/microhook-archive-lead.html in there. Now I have an archive page built around years and months. Good stuff.

ISO dates are back

Got ISO dates working on my blog again, just like the old days. Had to dig into microhooks and Hugo’s recommendation (hat-tip goes to Claude for making those more understandable). Turns out that “2006” is not just some random year value that can be replaced. Don’t worry, I’ll tell you why anyway.

What makes a good blog good?

Up early (again) with coffee in hand, pondering what separates the blogs I actually enjoy from the noise. It’s not the platform—Micro.blog, Bear, whatever—or perfect grammar. At the end, it’s the human showing through. What do you think?

My blog on Manue’s Prople and Blogs today!

manuelmoreale.com/interview…

It’s 4 AM, and I’m up again. I really need this insomnia to go away.

Apple Watch after 2 years

Apple Watch, year 2: excellent jogging buddy, sleep-shaming device, and an overpriced vibrating meditation reminder bracelet. MS Teams got banned immediately. But it’s not all bad.

Apple Mail - highlight a part of an email, and it shows up in your reply. Email threads are sorted by chronological order.

Outlook - Highlighting does nothing. Copy-paste automatically defaults to the original fonts and style, not that of the email. Email threads (“conversations”) have two timelines in one: chronological and reverse-chronological.

I’m starting to think they have a position for “torturer” they hire for now and then.

We've built an AI so good it will kill us with its kindness

I tried Claude like everyone else, and came back terrified because I found the best teacher I could have hoped for.

More signs of spring 📷

Auto-generated description: Tall trees with blooming white flowers reach towards a clear blue sky. Auto-generated description: A tall tree with ivy climbing its trunk stands among leafless trees against a bright blue sky. Auto-generated description: Branches of trees adorned with white blossoms are set against a bright blue sky.

I took a walk and listened to the birds for a moment. It’s always nice to get chances like this in Manhattan:

Auto-generated description: Sunlit stone steps wind through a lush, wooded area under a clear blue sky.

Fighting with bots. I’m not provided the option I need (work/enterprise-related shipment, and the only option on the website or on the phone is for home service.) The bot doesn’t route me to a human, no matter what I try. Some service.

One of my automatic tasks for my Mac is to delete my downloads folder every weekend. At the start, I was worried this would cause me to lose important files, but I have backups and it actually made me more purposeful with how I save them.

What are some of yours?

So I fixed my blog again

Spent days debugging my blog’s broken microhooks, only to find Hugo was using the “default” file instead of my custom one. Turns out, a flipped template and a missing dash in a title were the culprits. Shoutout to Qwen3-Coder for the assist. AI can be a great tutor. My blog matters more than I thought.