Hi HN,I’ve been working on Prostir, a simple productivity app that helps you plan your day, capture thoughts, and build habits — without feeling overwhelmed.The idea came from personal frustration: most tools either try to do everything or feel too rigid. I wanted something calmer — a space to think, not just check things off. So I built Prostir using Swift + SwiftUI, with focus on speed, minimalism, and offline support.I’ve been building Prostir, a mindful productivity app that helps you organize your day, jot down thoughts, and build healthy habits — without getting overwhelmed.It’s already available on the App Store and has gone through a few iterations based on early feedback. The goal has always been the same: to create a calm and simple space for planning, especially for people who find traditional task managers too rigid or stressful.Prostir is free to try, and if anyone here wants to explore the premium features, I made a promo code specifically for the HN community:
 Use promo code in appstore: &quot;HACKERNEWS&quot; 
Prostir: <a href="https:&#x2F;&#x2F;apps.apple.com&#x2F;us&#x2F;app&#x2F;daily-planner-prostir&#x2F;id6535696029">https:&#x2F;&#x2F;apps.apple.com&#x2F;us&#x2F;app&#x2F;daily-planner-prostir&#x2F;id653569...</a>Would really appreciate your thoughts or impressions. Always learning from how others approach building focused, human-friendly apps.Would love your feedback or ideas — especially if you’ve built something similar or tried calming productivity tools before.Thanks for checking it out!

Show HN: Prostir – A calm productivity app to plan without burnout

Show HN: An accurate implementation of Pomodoro Technique

Over the past few months, a student and I set out to explore the impact of vibe coding on web development. Somewhere along the way, we ended up building Daf·thunk, a visual workflow editor. It leverages Cloudflare’s fantastic infrastructure (Workers, D1, KV, Workflows, AI, etc.) to create surprisingly robust workflows that can be triggered manually, via HTTP requests, by email, or on a schedule.During development, we primarily used Cursor with its agent and tab modes, alongside Claude Sonnet 3.5, 3.7, 4, and Gemini 2.5 Pro. Occasionally, we switched to MAX Mode when working on or reviewing more complex changes. We tried to regularly refine our Cursor rules and began applying specific rules to different parts of the codebase (backend, frontend, database, etc.). We also indexed documentation and used it extensively in prompts. For large refactors, we often referenced previous commits to reapply patterns elsewhere in the code.Overall, we feel that prompting for small, incremental, and easy-to-review changes scales well when coding with LLMs and the results are really impressive. In this regard, Andrej Karpathy’s talk “Software Is Changing (Again)” resonates deeply. John Ousterhout’s concept of deep modules has also been a useful mental model. Our Cursor rules ask for simple APIs that hide rich internal logic, and avoid wide interfaces that mirror implementation details.Aside from frequent commits, we didn’t document our process much, as we were mainly exploring and building intuition around what works and what doesn’t. Since we chose to trust the LLM more than usual, we’ve released everything under an open source license and without warranty. As models improved, our reliance on unit tests dropped, which will probably come back to haunt us soon... contributions are welcome ;)GitHub repository:
<a href="https:&#x2F;&#x2F;github.com&#x2F;dafthunk-com&#x2F;dafthunk">https:&#x2F;&#x2F;github.com&#x2F;dafthunk-com&#x2F;dafthunk</a>Product Hunt page:
<a href="https:&#x2F;&#x2F;www.producthunt.com&#x2F;products&#x2F;dafthunk?launch=dafthunk" rel="nofollow">https:&#x2F;&#x2F;www.producthunt.com&#x2F;products&#x2F;dafthunk?launch=dafthun...</a>

Show HN: Daf·thunk – open-source Editor for Prototyping Workflows on Cloudflare

chatGPT image-1 is great but the ratio is so weird and do not respect 16:9 for example.So i made this free tool to get perfect ratio without cropping.Hope it is useful for some. Let me know

Show HN: Free tool to resize ChatGPT images to perfect 16:9

A year ago, I built this package but never shared it with any community. Sharing here in case this helps you in the future.However, if you like, don&#x27;t forget to give a star and drop your feedback

Show HN: Promise.allSettled Alternative with Concurrency Support

News-Hook is a tool for setting up webhooks on any real-world, using just prompts.You can write prompts like:“Inform me when [movie] gets a Rotten Tomatoes score”
“Tell me when [rumor] is confirmed or denied”
“Alert me on any tariffs news between USA and Brazil”Here’s how it works:-You write a prompt saying what you want to be informed about.
-You set up a webhook as always, with URL and payload.
-Done.There is a full API in case you wanna do this programatically
We got Swagger-UI open in case you wanna seeWould love feedback, use cases I haven’t thought of, and critiques on design or security.Thanks!

Show HN: News-Hook lets you prompt alerts to stay informed on anything

Hey HN!I’ve been working on Obelis AI — it’s like having a DevOps engineer on your team, powered by AI agents (real ones, not just a fancy LLM wrapper).The idea came from seeing how early-stage teams often push off infra decisions until it’s too late. Migrations are always an option, but they get painful fast as things scale.Platforms like Vercel and Firebase are great to get going, but they tend to get expensive, and you don’t really own the infra. We try to give you the same simplicity — but everything runs on your own cloud (AWS for the timebeing), with full control.We&#x27;re in beta now — would love feedback, questions, or any brutally honest thoughts. Appreciate it!

Show HN: DevOps made easier for startups

Hi HN,I’m Harrison, co-founder of Listed. Today we&#x27;re launching our agentic platform to help your business win in the new age of AI.You can try the platform here: <a href="https:&#x2F;&#x2F;getlisted.ai" rel="nofollow">https:&#x2F;&#x2F;getlisted.ai</a>
And watch the launch video here: <a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=MJUPo6H78z8" rel="nofollow">https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=MJUPo6H78z8</a>The idea for this came from pure frustration. I asked ChatGPT about my own company and it hallucinated, inventing features and getting basic facts wrong. I realized there was no mechanism for a business to provide a verified source of truth to these models.This problem is now existential. With Google&#x27;s AI Overviews and the rise of answer engines, your website&#x27;s unstructured HTML is a poor source for the rich, nuanced context that LLMs need. This leads to an army of AI bots from OpenAI, Google, Perplexity, etc., scraping your site, getting it wrong, and permanently baking those errors into their models.So, we built Listed.The simplest analogy is it&#x27;s like Cursor, but for context. Instead of an AI helping you write code, our agent helps you build the comprehensive, structured context that allows LLMs to represent your business accurately and favorably.Here’s how our agentic system works:Automated Context Building: When you sign up, our agent scrapes your existing website to build a first draft of your AI Listing. It structures the data and identifies weak spots.Intelligent Workflows: Based on ongoing analytics, the agent initiates simple, chat-based workflows to help you enrich your listing and improve its accuracy and ranking potential.Performance Analytics &amp; Feedback Loop: The agent constantly measures your AI Ranking (discoverability) and Recall Accuracy across all major models (GPT-4o, Claude 3, Gemini, etc.). This data feeds back into the system, generating new workflows to continuously improve your performance.The Connection: Your AI Listing is a hosted service. You add a simple code snippet to your website. When AI crawlers visit, this acts as a signpost, essentially &quot;prompt injecting&quot; and directing them to consume your clean, structured, AI-optimized data feed instead of trying to parse your messy site.The goal is to give every business an active role in the AI ecosystem. You provide the clean, verified data that AI companies desperately need, and in return, you get to control your narrative and rank higher in their answers.We are launching our free tier today. We’d love for you to try it out and hear your feedback.You can get started here: <a href="https:&#x2F;&#x2F;getlisted.ai" rel="nofollow">https:&#x2F;&#x2F;getlisted.ai</a>I&#x27;ll be here all day answering questions. Thanks!

Show HN: Listed – An agentic platform to rank your business on AI

Hello, everyone.Three months ago, I started building my first GoLang package, VarMQ. In this short period, I gained a significant amount of traction from the community, and it resulted in over 140 stars on GitHub.A week ago, to check how it performs, I ran benchmarks with a similar package, Pond, which has been managed for years.Pond is not a message queue, though; however, it can be compared to VarmQ since it shares some similarities, except for its storage system.From the benchmarks, I got surprising results, which are that Varmq is taking 50%+ less memory than Pond and also has good performance metrics in IO tasks. which makes me glad and willing to share with everyone here.Benchmark chart site: <a href="https:&#x2F;&#x2F;varmq-benchmarks.netlify.app" rel="nofollow">https:&#x2F;&#x2F;varmq-benchmarks.netlify.app</a>Overall, I would like to increase the adoption of this package and am willing to hear feedback from you.Regardless, if you like this project, don&#x27;t forget to hit the Star button. It helps the project grow.Thank you!

Show HN: A Storage-Agnostic Golang Message Queue

Three months ago someone posted a site that showed time in corporate [1]The most interesting comment to me was about the National Retail Federation 4-5-4 calendar. That calendar was hard to understand, so I made a visualization of it. I also implemented the other calendar types people were asking for.I did find the SEC dataset of 10032 publicly traded companies [2][3], but have not finished implementing a search for symbol yet.[1] <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=43162340">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=43162340</a>[2] <a href="https:&#x2F;&#x2F;www.sec.gov&#x2F;data-research&#x2F;sec-markets-data&#x2F;financial-statement-data-sets" rel="nofollow">https:&#x2F;&#x2F;www.sec.gov&#x2F;data-research&#x2F;sec-markets-data&#x2F;financial...</a> (SUB data set, field `fye`)[3] <a href="https:&#x2F;&#x2F;www.sec.gov&#x2F;files&#x2F;company_tickers.json" rel="nofollow">https:&#x2F;&#x2F;www.sec.gov&#x2F;files&#x2F;company_tickers.json</a>

Show HN: What time is it in Corporate

I built a tracker that automatically scrapes the White House website for new executive orders and uses GPT-4 to generate plain-English summaries.The system runs daily, finds new orders, feeds the full legal text to ChatGPT for summarization and auto-categorization, then generates individual pages and updates the main index. It even creates custom Open Graph images for social sharing.Currently tracking 158+ orders with automatic updates as new ones are signed.Features:
- AI summaries of all executive orders in plain English
- Auto-categorization by policy area (immigration, trade, AI, etc.)
- Search by keyword, date, or category
- Completely neutral
- Individual pages for each order with full text
- Auto-generated OG imagesI got tired of reading dense legal text to understand what&#x27;s actually being signed. The AI does the heavy lifting of parsing government language into readable summaries.Link: <a href="https:&#x2F;&#x2F;tonygaeta.com&#x2F;labs&#x2F;executive-orders" rel="nofollow">https:&#x2F;&#x2F;tonygaeta.com&#x2F;labs&#x2F;executive-orders</a>Tech: Next.js&#x2F;Tailwind frontend, Python scraper with BeautifulSoup, GPT-4 for summaries, automated OG image generation via headless chrome.

Show HN: AI-powered tracker of Trump executive orders

We built Chisel to make GPU kernel profiling hardware-free. It lets you run chisel profile kernel.cu and get full Nsight&#x2F;Ncompute or rocprofv3 reports without a GPU needed.It spins up remote H100, L40S, or MI300X machines (via DigitalOcean for now, but gonna expand backends soon), runs your code, and gives you back detailed traces (kernel timings, memory transfers, API calls, etc). Everything is CLI-based and designed for iterative dev—profiling takes \~1–2 minutes per run.For example:# Profile a PyTorch training script on H100 with Nsight Systems
chisel profile --nsys train.py# Profile a HIP kernel on MI300X with system trace
chisel profile --rocprofv3=&quot;--sys-trace&quot; matrix_add.cppRepo: <a href="https:&#x2F;&#x2F;github.com&#x2F;Herdora&#x2F;chisel">https:&#x2F;&#x2F;github.com&#x2F;Herdora&#x2F;chisel</a>
PyPI: pip install chisel-cliWould love feedback! especially from anyone building custom kernels, ML layers, or low-level GPU ops.

Show HN: Chisel – Profile GPU Kernels Without a GPU (Nvidia and AMD)

typo in the url: <a href="https:&#x2F;&#x2F;vibeflirting.com&#x2F;" rel="nofollow">https:&#x2F;&#x2F;vibeflirting.com&#x2F;</a>Hey HN,
After years of false starts learning to code (pure math background, but got stuck in tutorial hell), I finally shipped something real with Claude Code as my patient tutor.The problem: Most guys have terrible dating profile photos. Poor lighting, bad angles, unflattering composition. Men generally don&#x27;t like taking pics, so our photos suck.My solution: Built a pipeline that takes uploaded photos and enhances them specifically for dating apps by tying AI tools together. Trying different things but had the most success with training flux and using enhancor or kontext for skin texture (and midjourney for minor editing).Tech stack: Vanilla HTML&#x2F;CSS&#x2F;JS frontend, Netlify Functions backend, Cloudinary for storage, Google Sheets as database. No frameworks - kept it simple to actually finish something.(I also tried Tailwind with CSS but that eventually caused build conflicts, so I removed it and wrote everything in pure CSS. Lesson learned.)The learning curve was steep, but Claude Code answered literally hundreds of basic questions. I used todo folders to save context between sessions and screenshot folders for debugging help. Some other things too but with the pace of innovation, many of those are already outdated.Took about two months to get it all working. Mostly just happy to have shipped something and &quot;vibelearned.&quot; Hoping that by shipping and learning incrementally, I can tackle more complex projects.All advice for this new indie hacker&#x2F;newbie coder is welcome.RJ

Show HN: Vibeflirting turns bad selfies into good dating pics for Tinder

Show HN: Zizmor, static analysis for GitHub Actions

Chat with multiple AI models once and compare the results to pick the best one. This should help you with your research as different AI model can give you different answers and some might be better than others.

Show HN: Chat with ChatGPT, Claude, Grok, Gemini and Llama on One UI

Hi HN, Igor here, one of the engineers behind this project.High-quality voice APIs are usually either expensive, slow, or both. Cheaper and faster solutions very often lack realism. We decided to build Inworld TTS to bridge this gap.We just released two multilingual models. Our small model, named TTS-1, is on par with SOTA models quality-wise given objective metrics WER&#x2F;SIM&#x2F;DNSMOS. A larger model, TTS-1-Max, is even better. It can produce more nuanced speech and has ~3.5% better WER across all 11 supported languages averaged. Both models also support markup tags (e.g. prepend &quot;[happy]&quot; to the text to make the generation more enthusiastic, etc).The models are built with LLaMA 1B and 8B being the SpeechLM backbones for TTS-1 and TTS-1-Max respectively. We up-trained both models on a mixture of text and audio, then finetuned on text-audio pairs and polished final checkpoints with GRPO on a small high-quality dataset. Our Speech Lab team (4 MLEs) started to work on collecting audio data around late February and exploring different audio codec architectures. We got inspired by the simplicity of the single vector quantization Xcodec2 neural audio codec architecture used and decided to use a similar idea. We started the training early April. Once codec was ready, SpeechLMs’ training took another month and a half. We finished mid-June, all - using 32 H100 GPUs.To make models real-time ready during serving, we collaborated with Modular to migrate from vanilla vLLM solution to Mojo- written MAX server. Our bet of keeping serving architecture as simple as possible played out well: both models turned out to be really fast. TTS-1, which can be accessed via streaming API, has ~500ms p90 latency for returning the first ~2 seconds of audio. The pricing is simple, pay $5&#x2F;1M characters. A larger model’s API access will be opened soon. We’ll share more details about serving performance optimizations made in the coming weeks.We are also about to release all the training, modeling, and benchmarking code on GitHub to be transparent about how we made it. This repo is very flexible and can easily be adjusted to train an arbitrary neural net, but we’ll release the code with the focus on speech modeling. By the way, we’ve used PyTorch Lightning as the framework for multi-node&#x2F;multi-GPU training as it proved to be very easy-to-use and reliable.--Check the TTS out at <a href="https:&#x2F;&#x2F;inworld.ai&#x2F;tts" rel="nofollow">https:&#x2F;&#x2F;inworld.ai&#x2F;tts</a>Happy to answer any questions you have!

Show HN: Inworld TTS – high-quality, affordable, and low-latency TTS

With the Mozilla Pocket shutdown coming up in about two weeks, I thought I&#x27;d share this quick tool to migrate to linkding in case it&#x27;s helpful to others.After reviewing self-hosted options to Pocket, I decided linkding has the best combination of features. (The creator&#x2F;author of linkding has done a great job -- however, I plan to eventually create a new tool that is based on linkding but adds some new features that the author has indicated he doesn&#x27;t want to include [I’m currently using a fork, but I want to expand on it further].)HN thread about shutdown announcement: <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=44063662">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=44063662</a>Mozilla announcement: <a href="https:&#x2F;&#x2F;support.mozilla.org&#x2F;en-US&#x2F;kb&#x2F;future-of-pocket" rel="nofollow">https:&#x2F;&#x2F;support.mozilla.org&#x2F;en-US&#x2F;kb&#x2F;future-of-pocket</a>linkding: <a href="https:&#x2F;&#x2F;linkding.link&#x2F;" rel="nofollow">https:&#x2F;&#x2F;linkding.link&#x2F;</a>Note that Pocket is shutting down July 8, 2025, but the export service will remain available until October 8, 2025.[edit] fix typo in title &amp; formatting