Show HN за 3 июня 2026 г.

NoSuggest is a quiet act of resistance against YouTube algorithms always trying to pull you into a loop of unlimited videos in turn into unlimited screen time. With unending side cards of videos, auto-play, what's next suggestions, YouTube shorts and notifications, users will be doom scrolling for many hours in a day.

I faced the same problem. Acknowledging that, not all content in YouTube is bad. There are educational videos, genuine news contents without political bias which is very hard to find outside YouTube and many other good relaxing, entertainment stuff.

NoSuggest lets you only follow the YouTube channels you like and removes all types of recommendation YouTube has. So you don't waste time on watching things which you never wanted to watch anyways.

UI is very simple. You add your favourite channels in "Channels" tab and latest 5 videos per channel excluding shorts would appear in "Feed" tab. "Search" tab is to search for specific videos to watch and "Saved" tab is to bookmark any video you want to watch later. Intention of NoSuggest is to provide whatever is necessary to extract whats good from YouTube all inside NoSuggest and leave out bad parts.

NoSuggest works in any devices. Install it as an app (PWA) in android and iPhone, or simply open in browser in laptops. No sign-in, no account creation or no card details. NoSuggest won't even ask your name. Total privacy for the users.

Parents can add the channels and save some educational videos and lock it with the pin for kids mode. Kids won't be able access unwanted additive contents inside NoSuggest.

Completely free, no string attached. Source available in Github through NoSuggest website.

I would love genuine feedback. Thank you very much for your attention on this matter.

It's our new text-to-image model: a 9.3B single-stream diffusion transformer trained entirely from scratch.

We focused heavily on controllability through structured JSON prompts, with strong text rendering, spatial awareness through bounding box guidance, and color palette control.

It has the best text rendering of any open-weight model we've tested so far, and the NF4 quantized checkpoint runs on a single 24GB GPU.

For more technical details and examples see our blog post: https://ideogram.ai/blog/ideogram-4.0/

We will be happy to answer any questions :)

Turn complex PDFs into clean Markdown that people can review and AI tools can use.

Hello.

After hearing of the problematic LLM commits in rsync, I made a fork of rsync. I decided to fork it off release 3.4.1, since I heard that's the last release without the LLM code.

Hi HN,

I built Aura-IDE, a native desktop LLM coding harness for AI assisted software engineering.

The idea is not just “chat with your codebase.” Aura wraps models in a structured engineering loop:

repo awareness → Planner spec → Worker execution → surgical edits → validation → recovery → final receipt

The Planner reads the project and writes a spec. The Worker executes that spec with filesystem tools, diff approval, terminal validation, and recovery behavior. The goal is to make ordinary models produce better code by giving them better process, context, and tools.

The unusual part: Aura has been heavily dogfooded on itself. Roughly 98% of the codebase was generated, edited, or refined through Aura’s own Planner/Worker workflow under human direction. During May, visible DeepSeek usage crossed 1.1B tokens and nearly 30k API requests while improving Aura’s reliability, edit mechanics, UI, updater flow, README, demo flow, and product polish.

It is built with Python and PySide6. It supports DeepSeek, OpenAI, Anthropic, Gemini, OpenRouter, and CLI backends. It can run carefully with diff review, or faster with auto-dispatch/auto-approve.

I’d love feedback from people building coding agents, IDEs, or local/dev tools.

I was in a long-distance relationship for some time and it is hard. We love each other a lot but being in different countries, even different timezones for a longer time is quite a burden if you want to build a great relationship. We made it work (we are married now!) but it took a lot of work to stay and feel connected while we were not physically together. We tried quite a few things during that time, regular video calls, lots of text messages and even couple apps like Agape. You still struggle during those darker moments when fate makes your life a little more difficult. One day I had a good idea though that really helped us: I quite enjoy listening to podcasts so I just took my phone and started recording myself while I talked about some of the greatest moments we had together. Then I uploaded it to my own Nextcloud and shared it with my (now) wife. She loved it! Her own podcast full of joyful moments. Now there are a bunch of episodes already and she was able to listen to them any time - guaranteed to get some great memories in my own voice.

This really helped us and I am sure there are a lot of others out there struggling through very similar situations. That's why we started building a more professional tool for this: PersonalPod. I am the typical software engineer - I really enjoy building stuff (other than the UI) and I keep coming up with more ideas that will go into this project. This time we decided to make it available to others as well though before I build forever. Probably you actually need different things than I would think...

Have fun trying it out at https://personal-pod.com/ and please leave some feedback (here or directly on the website via the feedback tool). I will read them all I promise and try to improve things as much as I can.

You can already share podcasts with a small group of people (or just one if you want) or make a partner podcast where both of you can upload episodes.

as a big TUI user, i absolutely enjoy hunk https://github.com/modem-dev/hunk. tho when it comes to reading diffs i still prefer GUI somehow

made a local version of @pierrecomputer's diffshub.com for reviewing codes, with jj support and review copying to directly send to agents :D

huge shoutout to @bentlegen for the inspiration on agent review, and pierre team for the amazing work!

try it with `npx @baggiiiie@yadiff $GitRef/$JJRevision`

beyond local diff viewing, it supports GitHub PR fetching too. so you can test it on your PR, or test it on bun's rust rewrite PR `npx @baggiiiie@yadiff https://github.com/oven-sh/bun/pull/30412`

any feedback is appreciated!

Hello all. My background is in blockchain protocol engineering. The company I was at wound down, which left me with a lot of time, so I'd been pushing frontier models on hard distributed-systems problems and was impressed enough to try building something autonomous end-to-end.. a micro-SaaS factory. A few months in I realized the runtime underneath was more interesting than the product on top of it, so I extracted it. That's Swarm. A lot of the design comes straight from blockchain engineering: I wanted something purely async and event-driven, where state only moves through committed, ordered transitions. The one decision everything hangs on is that the LLM does not run the system. Agents reason in scoped sessions and emit events; deterministic code, never the model, decides what each result changes. I think this enables some interesting workflows. Curious to know what you think of it!

hi hn! Title basically covers it. I made Nib over the course of about a year and a half. It's a collaborative font editor, kind of like figma but for type design instead of UI design. Also has AI features for generating glyphs, versioned export, font import, advanced bézier curve editing features...and cursor chat. Among other features. Lemme know if you find any bugs!

Hey HN, we’re Soami, David, and Achyut, co-founders of Extract. Extract parses documents into structured data (text, tables, and figures). Teams use it for RAG, feeding llms, and populating databases and forms. Today we’re launching our first OCR model, now used in Extract.

You can try some examples here or upload your own (no signup required) to test it out: https://extract.page/demo

We built Extract out of YouLearn, where we were processing 70m+ pages and slow parsing was the bottleneck. We started with a purely algorithmic pipeline that pulled native text straight from the document and only ran OCR on pages that needed it. It was cheap and fast, but once we put it in front of our Extract customers and their hardest documents, it hit an accuracy ceiling. We wanted to keep the speed and cost while improving accuracy, so we trained our own VLM for the cases that broke. It also provides element level bboxes, so each result points back to its exact place on the page. That took one customer from 71% to 92% text accuracy in under a week, at the same speed and cost. We can do this because of our synthetic data generation pipeline that recreates the messy, real-world documents the model gets wrong, so we can retrain on those exact cases without having to hand-label data.

To see how this holds up against other providers, we benchmarked Extract against AWS Textract, Extend, Reducto, LlamaParse, and Unstructured on 130 human labeled pages from difficult real-world documents. Extract is #1 on text accuracy (81.9%) and word-overlap F1 (84.5%), second on grounded accuracy, and competitive on layout IoU, while running at least 2x faster than every parser we tested.

Here are the benchmarks: https://extract.page/bench

Extract is $3 per 1000 pages and about 5x cheaper than AWS Textract (layout + table enabled). To see how it performs on your own docs, feel free to send us a few and we’ll run a benchmark on them. We’ll get back to you with the results in a few days once we receive the docs: https://cal.com/team/youlearnai/extract-intro

Thanks for reading this post! It's our first version of the model and we're shipping further improvements to handwritten, multilingual, and table-heavy documents. We know there are documents it won't handle well yet. If you have one, we'd love to see it.