prtmnth

Am I missing something? Nextjs static export would have solved this I feel. And hosting it on Cloudflare Pages and put the CDN in front and you're done.

Having said that, sure perhaps Astro would be lighter than Nextjs.

I had to update because the previous OpenCode 2.0.3 I had installed stopped working suddenly. I love the idea behind OpenCode but compared to managing Claude and Codex, it's such a pain and requires continuous debugging or managing.

It's super tempting to work from my phone when I'm not at my desk. But every time I do that, let's say while lounging or during a walk, I realise it takes me away from being in the moment or experience of what I'm doing.

If I'm going on a walk, I like the complete disconnect, looking at nature and the mountains and goats and sheep on the roads.

If I'm lounging, I like being at rest and not active thinking with my brain.

Even passive thinking while speaking to an AI agent in the background breaks this flow state of being at rest and relaxation.

So for the time being, at the risk of being "less productive", I am happy to let go of speaking to my agents and getting things done while I am not working.

However, I do enjoy the benefits. I'm not a luddite. If there's a critical issue and I am notified and with a fix that's awaiting my approval, I'm glad that AI and phone can enable handling that situation without me knowing about it mid-hike and unable to do anything about it. So I'll still continue experimenting with working from the phone. But consciously, I want to design the experience for handing exceptions, and not make it an extension of me being at work at my desk.

This is amazing! We are using Box (now Boat dev) and yours is the first offering that looked even more economical. Will be exploring. Cheers and all the best

I've found Vite + React SPA ( Tanstack Router/Query) with a Nodejs/Fast API to work so simply and so easy to reason about. The entire SSR and caching etc. thing adds a lot of mental overhead.

Not to mention this stack can be deployed literally anywhere with great ease.

This news made me rethink about building on Bun. It's a bit iffy now.

I switched to Claude. Opus 5.5 was causing fomo anyway. This was the trigger. I admit I have zero loyalty

Using Tailscale in exactly the same way. It has been such a blessing!

Funnily I find the OP's thoughts to very much define what a main character is.

An NPC is given a role and identify and they are bound by it forever.

The main character chooses what they want. Whatever that is. But they exercise agency and their own will to choose their reality and what they wish to do and not do. What they wish to care and not care about.

The OP is not an NPC. I believe the OP wants to live a less stressful life. The OP is pulled towards a simple life that does not involve flashy things or high status (this is my interpretation).

Which in today's status signal fueled world may feel like being an NPC. But in my opinion, it is the exact opposite.

Through OpenRouter you can get Sol for $2 input / $10 output which makes it a really attractive choice amongst frontier models.

I cancelled my Claude subscription to give Codex a proper shot for a month. 1 week in, I'm more than happy. Sol is fast and capable. Luna as subagents for execution are very cheap usage wise (and API price wise too).

I think more than anything else, I don't get a headache conversing with Sol. That alone is enough reason for me to stick to Codex.

Experimenting with adding open source models to the mix to get more execution done while using Sol as the brain.

Online chatter seems to suggest it's either a GLM-5.3 Flash or a Microsoft model. I am leaning a flash model from GLM aiming to disrupt DeepSeek v4 Flash

I think my only gripe with DeepSeek was the lack of vision capabilities. This is an awesome and much awaited update!

Codex usage feels exorbitantly high since today. They [0] are denying it, but the number of anecdotal users who decided to raise this as an issue (as a result it's trending on X) says otherwise.

[0] https://x.com/thsottiaux/status/2090675027670978569

One of my favorite travel activities is browsing supermarkets and in awe of the interesting products on the shelves. So much fun!

Missed opportunity to call it DeepCode

I am curious to understand who is this model targeted at?

Before auto mode came out, I had a script that ran before every permission request, it called Haiku with a prompt with a list of safe and unsafe command examples and asked it to classify as safe/unsafe and log it so I can review it later. It worked really well for me until auto mode came out, at which point I preferred the provider's built-in classifier versus maintaining my own.

I've been using auto mode ever since the feature was released. Apart from a very few occassions where the classifier blocked a safe command, I have faced no issues and continue using it as my default mode. It's great!

I'm sitting at my desk browsing my phone and it gets locked with the theft protection alert. This only started happening in recent weeks. It's super annoying :/

I love using the Twitter web app over the native app. And whenever I think of a product, I think of how I can build a fully usable web app (even when primary access is via phone).

New product added to the Google Graveyard

Eagerly awaiting their V4 Pro update!

Haha, this is so cool and can't believe I didn't think about it. I've been opening 2 tabs per HN link on my phone since forever!

Anecdotal, I've been running my usual building skills which keep the main agent as the orchestrator and every step from execution, review, bug fix happens via subagents. So far in my 2 days of use of Opus 5 in medium and high effort modes, I haven't not observed it avoid using subagents.

I have also not observed it using less subagents than usual in regular chats. However, that could be due to limited use over a small timeframe.

See projects.dev by Stripe. My best hunch is - Stripe wants to get in on the infrastructure/rails layers being built for AI. They aren't sure what that is going to be exactly (nobody does). OpenRouter seems like a safe bet.

My hunch is Google is trying to integrate a fast and relatively cheap AI across search and every other surface of their product suite. And for that objective, a model that can move faster while being accurate and cheap enough is more important to them than producing a frontier class heavyweight model.

Why not a private GitHub repo? It's hard to believe that the technical team could not write 1 skill which all non-technical teams could use to sync their skills to GitHub.

Same here. I'd love a 2x Max plan! More than enough usage for my needs.

https://www.preetamnath.com/ and https://www.preetamnath.com/blog

I wish to write more often

Thank you for open sourcing and sharing the GitHub repo. As someone actively trying to upgrade my skills when it comes to coding and building apps, this is really useful.