The best Hacker News stories from Show from the past day
Latest posts:
Show HN: Learn LLMs LeetCode Style
Show HN: Learn LLMs LeetCode Style
Show HN: A Raycast-compatible launcher for Linux
Hey HN!<p>I'm a huge fan of Raycast, but as a Linux user, I was always disappointed it wasn't available on my main OS. This summer, I decided to just build it myself. This project has the goal of being interoperable with Raycast itself, including a majority of the extensions.<p>It's built with Tauri and Rust on the backend, with a Svelte frontend. The biggest challenge was getting it to run existing Raycast extensions, which required building a custom React renderer as well as making a custom API.<p>I also wrote a quick post, which I hope to expand on in the future, about this project. You can find it here: <a href="https://byteatatime.dev/posts/recreating-raycast" rel="nofollow">https://byteatatime.dev/posts/recreating-raycast</a><p>The project is still very rough, but I'm sharing it now to get any feedback you may have!
Show HN: A Raycast-compatible launcher for Linux
Hey HN!<p>I'm a huge fan of Raycast, but as a Linux user, I was always disappointed it wasn't available on my main OS. This summer, I decided to just build it myself. This project has the goal of being interoperable with Raycast itself, including a majority of the extensions.<p>It's built with Tauri and Rust on the backend, with a Svelte frontend. The biggest challenge was getting it to run existing Raycast extensions, which required building a custom React renderer as well as making a custom API.<p>I also wrote a quick post, which I hope to expand on in the future, about this project. You can find it here: <a href="https://byteatatime.dev/posts/recreating-raycast" rel="nofollow">https://byteatatime.dev/posts/recreating-raycast</a><p>The project is still very rough, but I'm sharing it now to get any feedback you may have!
Show HN: A Raycast-compatible launcher for Linux
Hey HN!<p>I'm a huge fan of Raycast, but as a Linux user, I was always disappointed it wasn't available on my main OS. This summer, I decided to just build it myself. This project has the goal of being interoperable with Raycast itself, including a majority of the extensions.<p>It's built with Tauri and Rust on the backend, with a Svelte frontend. The biggest challenge was getting it to run existing Raycast extensions, which required building a custom React renderer as well as making a custom API.<p>I also wrote a quick post, which I hope to expand on in the future, about this project. You can find it here: <a href="https://byteatatime.dev/posts/recreating-raycast" rel="nofollow">https://byteatatime.dev/posts/recreating-raycast</a><p>The project is still very rough, but I'm sharing it now to get any feedback you may have!
Show HN: RULER – Easily apply RL to any agent
Hey HN, Kyle here, one of the co-founders of OpenPipe.<p>Reinforcement learning is one of the best techniques for making agents more reliable, and has been widely adopted by frontier labs. However, adoption in the outside community has been slow because it's so hard to implement.<p>One of the biggest challenges when adapting RL to a new task is the need for a task-specific "reward function" (way of measuring success). This is often difficult to define, and requires either high-quality labeled data and/or significant domain expertise to generate.<p>RULER is a drop-in reward function that works across different tasks without any of that complexity.<p>It works by showing N trajectories to an LLM judge and asking it to rank them relative to each other. This sidesteps the calibration issues that plague most LLM-as-judge approaches. Combined with GRPO (which only cares about relative scores within groups), it just works (surprisingly well!).<p>We have a full writeup on the blog, including results on 4 production tasks. On all 4 tasks, small Qwen 2.5 models trained with RULER+GRPO beat the best prompted frontier model, despite being significantly smaller and cheaper to run. Surprisingly, they even beat models trained with hand-crafted reward functions on 3/4 tasks! <a href="https://openpipe.ai/blog/ruler">https://openpipe.ai/blog/ruler</a><p>Repo: <a href="https://github.com/OpenPipe/ART">https://github.com/OpenPipe/ART</a>
Show HN: RULER – Easily apply RL to any agent
Hey HN, Kyle here, one of the co-founders of OpenPipe.<p>Reinforcement learning is one of the best techniques for making agents more reliable, and has been widely adopted by frontier labs. However, adoption in the outside community has been slow because it's so hard to implement.<p>One of the biggest challenges when adapting RL to a new task is the need for a task-specific "reward function" (way of measuring success). This is often difficult to define, and requires either high-quality labeled data and/or significant domain expertise to generate.<p>RULER is a drop-in reward function that works across different tasks without any of that complexity.<p>It works by showing N trajectories to an LLM judge and asking it to rank them relative to each other. This sidesteps the calibration issues that plague most LLM-as-judge approaches. Combined with GRPO (which only cares about relative scores within groups), it just works (surprisingly well!).<p>We have a full writeup on the blog, including results on 4 production tasks. On all 4 tasks, small Qwen 2.5 models trained with RULER+GRPO beat the best prompted frontier model, despite being significantly smaller and cheaper to run. Surprisingly, they even beat models trained with hand-crafted reward functions on 3/4 tasks! <a href="https://openpipe.ai/blog/ruler">https://openpipe.ai/blog/ruler</a><p>Repo: <a href="https://github.com/OpenPipe/ART">https://github.com/OpenPipe/ART</a>
Show HN: DesignArena – crowdsourced benchmark for AI-generated UI/UX
I’ve been using AI to generate some repetitive frontend (guilty), and while most outputs felt vibe-coded, some results were surprisingly good. So I cleaned it up and made a ranking game out of it with friends, and you can check it out here: <a href="https://www.designarena.ai/vote" rel="nofollow">https://www.designarena.ai/vote</a><p>/vote: Your prompt will be answered by four random, anonymous models. You pick the one you prefer and crown the winner, tournament-style.<p>/leaderboard: See the current winning models, as dictated by voter preferences.<p>/play: Iterate quickly by seeing four models respond to the same input and pressing space to regenerate the results you don’t lock-in.<p>We were especially impressed with the quality of DeepSeek and Grok, and variance between categories (To judge by the results so far, OpenAI is very good for game dev, but seems to suck everywhere else).<p>We’ve learned a lot, and are curious to hear your comments and questions. Excited to make this better!
Show HN: DesignArena – crowdsourced benchmark for AI-generated UI/UX
I’ve been using AI to generate some repetitive frontend (guilty), and while most outputs felt vibe-coded, some results were surprisingly good. So I cleaned it up and made a ranking game out of it with friends, and you can check it out here: <a href="https://www.designarena.ai/vote" rel="nofollow">https://www.designarena.ai/vote</a><p>/vote: Your prompt will be answered by four random, anonymous models. You pick the one you prefer and crown the winner, tournament-style.<p>/leaderboard: See the current winning models, as dictated by voter preferences.<p>/play: Iterate quickly by seeing four models respond to the same input and pressing space to regenerate the results you don’t lock-in.<p>We were especially impressed with the quality of DeepSeek and Grok, and variance between categories (To judge by the results so far, OpenAI is very good for game dev, but seems to suck everywhere else).<p>We’ve learned a lot, and are curious to hear your comments and questions. Excited to make this better!
Show HN: BinaryRPC – Lightweight WebSocket-based RPC framework in modern C++
Hi HN,<p>I’m a recent CS graduate.
During the past few months I wrote BinaryRPC, an open-source RPC framework in modern C++20 focused on low-latency, binary WebSocket messaging.<p>Why I built it
* Wanted first-class session support, pluggable QoS levels and a simple middleware chain (global, specific, multi handler) without extra JSON/XML parsing.
* Easy developer experience<p>A quick feature list
* Binary WebSocket frames – minimal overhead
* Built-in session layer (login / reconnect / heartbeat)
* QoS1 / QoS2 with automatic ACK & retry
* Plugin system – rooms, msgpack, etc. can be added in one line
* Thread-safe core: RAII + folly<p>Still early (solo project), so any feedback on design, concurrency model or missing must-have features would help a lot.<p>Thanks for reading!<p>also see "Chat Server in 5 Minutes with BinaryRPC": <a href="https://medium.com/@efecanerdem0907/building-a-chat-server-in-5-minutes-with-binaryrpc-qos2-session-management-and-room-plugin-ccb66d722466" rel="nofollow">https://medium.com/@efecanerdem0907/building-a-chat-server-i...</a>
Show HN: BinaryRPC – Lightweight WebSocket-based RPC framework in modern C++
Hi HN,<p>I’m a recent CS graduate.
During the past few months I wrote BinaryRPC, an open-source RPC framework in modern C++20 focused on low-latency, binary WebSocket messaging.<p>Why I built it
* Wanted first-class session support, pluggable QoS levels and a simple middleware chain (global, specific, multi handler) without extra JSON/XML parsing.
* Easy developer experience<p>A quick feature list
* Binary WebSocket frames – minimal overhead
* Built-in session layer (login / reconnect / heartbeat)
* QoS1 / QoS2 with automatic ACK & retry
* Plugin system – rooms, msgpack, etc. can be added in one line
* Thread-safe core: RAII + folly<p>Still early (solo project), so any feedback on design, concurrency model or missing must-have features would help a lot.<p>Thanks for reading!<p>also see "Chat Server in 5 Minutes with BinaryRPC": <a href="https://medium.com/@efecanerdem0907/building-a-chat-server-in-5-minutes-with-binaryrpc-qos2-session-management-and-room-plugin-ccb66d722466" rel="nofollow">https://medium.com/@efecanerdem0907/building-a-chat-server-i...</a>
Show HN: I built a playground to showcase what Flux Kontext is good at
Hi HN,<p>After spending some time with the new `flux kontext dev` model, I realized its most powerful capabilities aren't immediately obvious. Many people might miss its true potential by just scratching the surface.<p>I went deep and curated a collection of what I think are its most interesting use cases – things like targeted text removal, subtle photo restoration, and creative style transfers.<p>I felt that simply writing about them wasn't enough. The best way to understand the value is to see it and try it for yourself.<p>That's why I built FluxKontextLab (<a href="https://fluxkontextlab.com" rel="nofollow">https://fluxkontextlab.com</a>).<p>On the site, I've presented these curated examples with before-and-after comparisons. More importantly, there's an interactive playground right there, so you can immediately test these ideas or your own prompts on your own images.<p>My goal is to share what this model is capable of beyond the basics.<p>It's still an early project. I'd love for you to take a look and share your thoughts or any cool results you generate.
Show HN: I built a playground to showcase what Flux Kontext is good at
Hi HN,<p>After spending some time with the new `flux kontext dev` model, I realized its most powerful capabilities aren't immediately obvious. Many people might miss its true potential by just scratching the surface.<p>I went deep and curated a collection of what I think are its most interesting use cases – things like targeted text removal, subtle photo restoration, and creative style transfers.<p>I felt that simply writing about them wasn't enough. The best way to understand the value is to see it and try it for yourself.<p>That's why I built FluxKontextLab (<a href="https://fluxkontextlab.com" rel="nofollow">https://fluxkontextlab.com</a>).<p>On the site, I've presented these curated examples with before-and-after comparisons. More importantly, there's an interactive playground right there, so you can immediately test these ideas or your own prompts on your own images.<p>My goal is to share what this model is capable of beyond the basics.<p>It's still an early project. I'd love for you to take a look and share your thoughts or any cool results you generate.
Show HN: Interactive pinout for the Raspberry Pi Pico 2
I've been trying to make accessible and beautiful GPIO pinouts since I started one for the Raspberry Pi in 2013 [1]. I've since given the Raspberry Pi Pico [2] and Pico 2 [3] microcontrollers the same treatment when they launched.<p>Recently I've updated these with a new "Upside-down" view to complement the rear view, giving a pinout in the right orientation to match your project.<p>The Pico sites are all hand-coded single HTML pages with supporting CSS and minimal JS. They are set up to optionally install as a "Desktop" web app. They also degrade into a somewhat usable table in lieu of CSS and use vector graphics (for the board itself) to be viewable and printable at any size.<p>Finally, hidden behind "Advanced" is a pinout of the test pads and special function pins!<p>[1] - <a href="https://web.archive.org/web/20130505194305/pi.gadgetoid.com/pinout" rel="nofollow">https://web.archive.org/web/20130505194305/pi.gadgetoid.com/...</a>
[2] - <a href="https://pico.pinout.xyz" rel="nofollow">https://pico.pinout.xyz</a>
[3] - <a href="https://pico2.pinout.xyz" rel="nofollow">https://pico2.pinout.xyz</a>
Show HN: Interactive pinout for the Raspberry Pi Pico 2
I've been trying to make accessible and beautiful GPIO pinouts since I started one for the Raspberry Pi in 2013 [1]. I've since given the Raspberry Pi Pico [2] and Pico 2 [3] microcontrollers the same treatment when they launched.<p>Recently I've updated these with a new "Upside-down" view to complement the rear view, giving a pinout in the right orientation to match your project.<p>The Pico sites are all hand-coded single HTML pages with supporting CSS and minimal JS. They are set up to optionally install as a "Desktop" web app. They also degrade into a somewhat usable table in lieu of CSS and use vector graphics (for the board itself) to be viewable and printable at any size.<p>Finally, hidden behind "Advanced" is a pinout of the test pads and special function pins!<p>[1] - <a href="https://web.archive.org/web/20130505194305/pi.gadgetoid.com/pinout" rel="nofollow">https://web.archive.org/web/20130505194305/pi.gadgetoid.com/...</a>
[2] - <a href="https://pico.pinout.xyz" rel="nofollow">https://pico.pinout.xyz</a>
[3] - <a href="https://pico2.pinout.xyz" rel="nofollow">https://pico2.pinout.xyz</a>
Show HN: Vibe Kanban – Kanban board to manage your AI coding agents
Hey HN! I'm Louis, one of the creators of Vibe Kanban.<p>We started working on this a few weeks ago. Personally, I was feeling pretty useless working synchronously with coding agents. The 2-5 minutes that they take to complete their work often led me to distraction and doomscrolling.<p>But there's plenty of productive work that we (human engineers) could be doing in that time, especially if we run coding agents in the background and parallelise them.<p>Vibe Kanban lets you effortlessly spin up multiple coding agents. While some agents handle tasks in the background, you can focus on planning future work or reviewing completed tasks.<p>After a few weeks of internal dog fooding and sharing it with friends, we've now open-sourced Vibe Kanban, and it's stable enough for day-to-day use.<p>I'd love to hear your feedback, feel free to open an issue on the github and we'll respond ASAP.
Show HN: Vibe Kanban – Kanban board to manage your AI coding agents
Hey HN! I'm Louis, one of the creators of Vibe Kanban.<p>We started working on this a few weeks ago. Personally, I was feeling pretty useless working synchronously with coding agents. The 2-5 minutes that they take to complete their work often led me to distraction and doomscrolling.<p>But there's plenty of productive work that we (human engineers) could be doing in that time, especially if we run coding agents in the background and parallelise them.<p>Vibe Kanban lets you effortlessly spin up multiple coding agents. While some agents handle tasks in the background, you can focus on planning future work or reviewing completed tasks.<p>After a few weeks of internal dog fooding and sharing it with friends, we've now open-sourced Vibe Kanban, and it's stable enough for day-to-day use.<p>I'd love to hear your feedback, feel free to open an issue on the github and we'll respond ASAP.
Show HN: Vibe Kanban – Kanban board to manage your AI coding agents
Hey HN! I'm Louis, one of the creators of Vibe Kanban.<p>We started working on this a few weeks ago. Personally, I was feeling pretty useless working synchronously with coding agents. The 2-5 minutes that they take to complete their work often led me to distraction and doomscrolling.<p>But there's plenty of productive work that we (human engineers) could be doing in that time, especially if we run coding agents in the background and parallelise them.<p>Vibe Kanban lets you effortlessly spin up multiple coding agents. While some agents handle tasks in the background, you can focus on planning future work or reviewing completed tasks.<p>After a few weeks of internal dog fooding and sharing it with friends, we've now open-sourced Vibe Kanban, and it's stable enough for day-to-day use.<p>I'd love to hear your feedback, feel free to open an issue on the github and we'll respond ASAP.
Show HN: Cactus – Ollama for Smartphones
Hey HN, Henry and Roman here - we've been building a cross-platform framework for deploying LLMs, VLMs, Embedding Models and TTS models locally on smartphones.<p>Ollama enables deploying LLMs models locally on laptops and edge severs, Cactus enables deploying on phones. Deploying directly on phones facilitates building AI apps and agents capable of phone use without breaking privacy, supports real-time inference with no latency, we have seen personalised RAG pipelines for users and more.<p>Apple and Google actively went into local AI models recently with the launch of Apple Foundation Frameworks and Google AI Edge respectively. However, both are platform-specific and only support specific models from the company. To this end, Cactus:<p>- Is available in Flutter, React-Native & Kotlin Multi-platform for cross-platform developers, since most apps are built with these today.<p>- Supports any GGUF model you can find on Huggingface; Qwen, Gemma, Llama, DeepSeek, Phi, Mistral, SmolLM, SmolVLM, InternVLM, Jan Nano etc.<p>- Accommodates from FP32 to as low as 2-bit quantized models, for better efficiency and less device strain.<p>- Have MCP tool-calls to make them performant, truly helpful (set reminder, gallery search, reply messages) and more.<p>- Fallback to big cloud models for complex, constrained or large-context tasks, ensuring robustness and high availability.<p>It's completely open source. Would love to have more people try it out and tell us how to make it great!<p>Repo: <a href="https://github.com/cactus-compute/cactus">https://github.com/cactus-compute/cactus</a>
Show HN: Cactus – Ollama for Smartphones
Hey HN, Henry and Roman here - we've been building a cross-platform framework for deploying LLMs, VLMs, Embedding Models and TTS models locally on smartphones.<p>Ollama enables deploying LLMs models locally on laptops and edge severs, Cactus enables deploying on phones. Deploying directly on phones facilitates building AI apps and agents capable of phone use without breaking privacy, supports real-time inference with no latency, we have seen personalised RAG pipelines for users and more.<p>Apple and Google actively went into local AI models recently with the launch of Apple Foundation Frameworks and Google AI Edge respectively. However, both are platform-specific and only support specific models from the company. To this end, Cactus:<p>- Is available in Flutter, React-Native & Kotlin Multi-platform for cross-platform developers, since most apps are built with these today.<p>- Supports any GGUF model you can find on Huggingface; Qwen, Gemma, Llama, DeepSeek, Phi, Mistral, SmolLM, SmolVLM, InternVLM, Jan Nano etc.<p>- Accommodates from FP32 to as low as 2-bit quantized models, for better efficiency and less device strain.<p>- Have MCP tool-calls to make them performant, truly helpful (set reminder, gallery search, reply messages) and more.<p>- Fallback to big cloud models for complex, constrained or large-context tasks, ensuring robustness and high availability.<p>It's completely open source. Would love to have more people try it out and tell us how to make it great!<p>Repo: <a href="https://github.com/cactus-compute/cactus">https://github.com/cactus-compute/cactus</a>