Artem Horobchenko

@johnnybuildr @TTrimoreau Yes, you're definitely right that most people now prefer a dark theme (me included) and get upset when they don't see one. In my comment, I was mainly pointing out that if the product itself isn't presented well, a dark theme won't make any difference.

@SSShken @TTrimoreau I respectfully disagree. I don't care for dark mode myself but I've had someone leave a 3 star review because they wanted dark mode, no other reason given. May depend on the niche but anything with a lot of screen time like a habit tracker then yes, dark mode should be built in

Builders, Does adding dark mode and light mode actually increase user retention?

@PaulFidika yep, best part - my worst session WAS "managed", continued from a summary. It grew right back past 100M

@SSShken interesting; larger context window size can be a footgun for the agent harness if it isn't managing its context well

Thoughts on GPT 5.6 Sol, after coding with Fable 5 for the past three weeks: - Codex needs a sub-agent view; hard to see what's going on. - Sessions are limited to a max of 3 sub-agents; kind of lame, but perhaps that keeps the chaos at bay. - Sol feels a lot faster than Fable 5 - The codex harness now rate limits by week only now; no more session limits. That's great. I'm on track to blow through my $200 Pro account's weekly allowance in a day though. - The new 'usage reset' burnable credit is a pretty cool bonus; I like it. - Sol seems on par with Fable 5 for intelligence; I need more time to feel this out though. - Sol keeps going for hours, whereas 5.5 used to stop every 20 mins for no reason; I would queue up a bunch of 'did you finish your issue? if not keep going' messages.

The flip side of that full 1M window: nothing forces you to compact, so the whole history rides along on every request. I parsed my Claude Code logs these days — one long session quietly ate 128M tokens, mostly re-reading its own context. Big window is a feature that bills like one

- Sol supports a 1 million token context window, but the codex harness limits it to only 257k, whereas Claude Code allows all its models to use their full 1 million context window. This means codex / Sol does much more frequent compacting.

@levelsio Is that Malta?

Sunmaxxxing today

@claudeai Will Fable be included?

Read more about how Claude for Teachers works: anthropic.com/news/claude-fo…

We're introducing Claude for Teachers: free access to premium Claude capabilities for verified K-12 educators in the US, with a library of teaching skills and a direct connection to evidence-based curricula, mapped to academic standards in all 50 states. claude.com/solutions/teac…

@marclou Crazy, next goal is 180

how to scare a customer

Wild detail: my fresh input was 75K tokens of those 236M. 0.03%. The rest is the machine re-reading itself. Built read-only, local, in one evening. If you want to run it on your own logs, say so. I'll open-source it.

Stop burning tokens. I just learned where mine actually go. I'm on the $100 Claude plan, writing prompts by hand, and already hitting limits. So I built a tiny parser for my Claude Code logs. Last 14 days: 236M tokens. I expected a caching problem. Nope: 99.9% cache hit. The real leak? Marathon sessions. The top one alone: 128M+ tokens, dragging the whole history behind it like a trailer. And that's manual me. When agents run agents, this multiplies. I'm building a tool that reads your AI usage and tells you what to fix. Finding #1. More coming.

@NicolasLecocqHQ Amazing! The font-size/weight of these things should be big enough to see them immediately

@SSShken I have 3, security, AI visibility and worldwide compliant I put them upfront

Day 6 of building in public. Goal: $10,000 MRR Current: $0 MRR 454 visitors on Amabrik in 30 days and 0 sales. I could blame the traffic, but it did its job. I keep pushing and keep rewriting until it converts. What is the one thing on a landing page that makes you pull out your card?

@levelsio @grok What can he do?

My Tesla Y 2025 windshield wiper fluid doesn't come out anymore, when I press it, it sprays it out of the bumper Seems like the hose is lose? How I fix this? I already removed the top part under the hood, but the problem isn't there, it's more below near the front bumper That part is fixed with some screws though, should I open it? THANKS

@arvidkahl They will stop building and start fighting

I want to run Clauded Code AND Codex on the same codebase at the same time, but I really don't want to have to deal with worktrees and provisioning db/redis/opensearch for each of them...

@ItsAlexhere0 Exactly!

@SSShken That's actually a really good example. Those tiny touches are easy to overlook but make a huge difference

Tell me one thing you can do that CLAUDE cannot do yet

@levelsio Just a huge monopoly...

How they want you to live - tiny windows to save energy - no AC to save the climate - vegan plant-based food to save the animals - work non-stop to pay tax for governments to waste it - fully medicated so you don't complain

@TTrimoreau Not yet, but i'm in the process of building the tool based on my own problem, and it's actually fantastic and that's probably how big guys generated revenue

Founders, Be honest are you using your own products?

@wickedguro I hope everyone has as much as he does

Postiz reached $155k MRR. $2m ARR getting closer 😵‍💫

@marclou I see that, and it motivates me to grow!

my new feed is full of indie hackers i follow, it's so nice 😊

@levelsio A real man doesn't even cry when chopping onions

You end up with nails smelling like garlic for the next 4 weeks and crying from cutting onions, literally no reason to not just buy pre-cut