🛠️ Hacker News Tools

[Other] Pseudo, a Common Lisp macro for pseudocode expressions

[Other] Show HN: The easiest accessibility (a11y) checker for VSCode

[Other] Show HN: Xorq – open compute catalog for AI Hi HN, Hussain and Dan from Xorq here.<p>After years of struggling with scaling compute that worked in notebooks but failed in production, we decided to do something about it. Data has standards like Iceberg and Delta. But compute is still a mess—trapped in notebooks, duplicated effort across teams, or baked into custom Airflow DAGs. We think of Xorq as the missing analog to Apache Iceberg, but for compute.<p>We’ve spent the last year building Xorq, an *compute catalog* that helps teams *reuse, ship, and observe* transformations, features, models, and pipelines across engines.<p>Xorq is built on:<p>- *Arrow Flight* (`do_exchange`) for high-speed data transport - *Ibis* for cross-engine expression trees, serialized to YAML - A portable UDF engine that compiles pipelines to SQL or Python - `uv` to make Python environments fully reproducible<p>Xorq features:<p>- pandas-style declarative transformations, backed by Ibis - Multi-engine execution (e.g., DuckDB, Snowflake) - UDFs as portable Flight endpoints - Serveable transforms by way of flight_udxf operator - Built-in caching and lineage tracking - Diff-able YAML artifacts, great for CI&#x2F;CD<p>Xorq use cases:<p>Since our last major release, it’s been exciting to see the first Xorq use-cases show up in the wild. All with *Python simplicity and SQL-scale performance*.<p>- Feature Stores (<a href="https:&#x2F;&#x2F;www.xorq.dev&#x2F;blog&#x2F;featurestore-to-featurehouse" rel="nofollow">https:&#x2F;&#x2F;www.xorq.dev&#x2F;blog&#x2F;featurestore-to-featurehouse</a>) - Semantic Layers (e.g. <a href="https:&#x2F;&#x2F;github.com&#x2F;boringdata&#x2F;boring-semantic-layer">https:&#x2F;&#x2F;github.com&#x2F;boringdata&#x2F;boring-semantic-layer</a>) - MCP + ML Integration (<a href="https:&#x2F;&#x2F;docs.xorq.dev&#x2F;vignettes&#x2F;mcp_flight_server" rel="nofollow">https:&#x2F;&#x2F;docs.xorq.dev&#x2F;vignettes&#x2F;mcp_flight_server</a>)<p>We’re open source and learning fast. Would love feedback on what’s useful or missing. Thanks in advance for trying it out!<p>Check out the demo of the Xorq CLI tool in action: <a href="https:&#x2F;&#x2F;asciinema.org&#x2F;a&#x2F;730484" rel="nofollow">https:&#x2F;&#x2F;asciinema.org&#x2F;a&#x2F;730484</a><p>---<p>Get Started<p>- Github: <a href="https:&#x2F;&#x2F;github.com&#x2F;xorq-labs&#x2F;xorq">https:&#x2F;&#x2F;github.com&#x2F;xorq-labs&#x2F;xorq</a> - Xorq docs: <a href="https:&#x2F;&#x2F;docs.xorq.dev&#x2F;" rel="nofollow">https:&#x2F;&#x2F;docs.xorq.dev&#x2F;</a> ---<p>Sneak peak - Xorq Compute Catalog UI Console:<p>Check out this interactive Claude demo showing how the Xorq compute catalog can be visualized to accelerate composition, reuse, and troubleshooting of AI compute: <a href="https:&#x2F;&#x2F;claude.ai&#x2F;public&#x2F;artifacts&#x2F;d2f00d2a-a3f9-4032-884e-d22f620a0ccf?fullscreen=true" rel="nofollow">https:&#x2F;&#x2F;claude.ai&#x2F;public&#x2F;artifacts&#x2F;d2f00d2a-a3f9-4032-884e-d...</a>

[Other] Show HN: ELF Injector The ELF Injector allows you to &quot;inject&quot; arbitrary-sized relocatable code chunks into ELF executables. The code chunks will run before the original entry point of the executable runs.<p>Included in the project are sample chunks as well as a step-by-step tutorial on how it works.<p>It&#x27;s a mix of C and assembly and currently runs on 32-bit ARM though it&#x27;s easy to port to other architectures.

[Other] Structuring large Clojure codebases with Biff

[Other] Show HN: Wush-Action – SSH into GitHub Actions over WireGuard

[Other] Show HN: Same prompt tested across Replit, Bolt, v0, Lovable and Raq.com Hi HN,<p>I built Raq.com – a platform that uses Claude Code to build working internal tools directly in the browser.<p>Claude Code is great at self correcting when given the right tools.<p>I&#x27;ve found that the popular web-based AI coding tools look great in demos but fail on real API integrations or require a lot of error back and forth. They don&#x27;t appear to do much research or self-correcting, likely to reduce spend. I wanted to see the current state of these tools, so I ran the same prompt on five platforms (Replit, Bolt, v0, Lovable, and Raq.com) to build a tool that requires 3 different APIs (Companies House, FinUK and OpenRouter) working together.<p>Four platforms produced broken prototypes or needed manual fixes. Raq.com delivered a complete working solution from a single prompt (that can be deployed to live with one click).<p>Full test with videos: <a href="https:&#x2F;&#x2F;raq.com&#x2F;real-world-test" rel="nofollow">https:&#x2F;&#x2F;raq.com&#x2F;real-world-test</a><p>We&#x27;re in early access (requires Claude Pro&#x2F;Max for free usage) - we&#x27;re looking non-coders who would like to build internal tools for their team.<p>Some technical info:<p>- Raq.com provisions isolated dev and prod Docker environments for each company (companyname.raq.com and companyname-dev.raq.com).<p>- The dev site includes a persistent terminal streamed to the browser, so the session continues even while tab is closed.<p>- CLAUDE.md file provides best practices, known pitfalls, and coding patterns for the Laravel + Filament stack.<p>- Self-Correction Loop: Claude can test and debug its own work. It has direct shell access to a custom script that bundles PHPUnit, syntax checks, and cache clearing. Plus a Playwright wrapper to check for errors and take screenshots.<p>- A single click runs a script that rsync&#x27;s the dev workspace to the prod container, runs migrations, and clears caches.

[Other] Show HN: Terminal-Bench-RL: Training Long-Horizon Terminal Agents with RL After training calculator agent via RL, I really wanted to go bigger! So I built RL infrastructure for training long-horizon terminal&#x2F;coding agents that scales from 2x A100s to 32x H100s (~$1M worth of compute!) Without any training, my 32B agent hit #19 on Terminal-Bench leaderboard, beating Stanford&#x27;s Terminus-Qwen3-235B-A22! With training... well, too expensive, but I bet the results would be good!<p>*What I did*:<p>- Created a Claude Code-inspired agent (system msg + tools)<p>- Built Docker-isolated GRPO training where each rollout gets its own container<p>- Developed a multi-agent synthetic data pipeline to generate &amp; validate training data with Opus-4<p>- Implemented a hybrid reward signal of unit test verifiers &amp; a behavioural LLM judge.<p>*Key results*:<p>- My untrained Qwen3-32B agent achieved 13.75% on Terminal-Bench (#19, beats Stanford&#x27;s Qwen3-235B MoE)<p>- I tested training to work stably on 32x H100s distributed across 4 bare metal nodes<p>- I created a mini-eval framework for LLM-judge performance. Sonnet-4 won.<p>- ~£30-50k needed for full training run of 1000 epochs (I could only afford testing )<p>*Technical details*:<p>- The synthetic dataset ranges from easy to extremely hard tasks. An example hard task&#x27;s prompt:<p>&quot;I found this mystery program at `&#x2F;app&#x2F;program` and I&#x27;m completely stumped. It&#x27;s a stripped binary, so I have no idea what it does or how to run it properly. The program seems to expect some specific input and then produces an output, but I can&#x27;t figure out what kind of input it needs. Could you help me figure out what this program requires?&quot;<p>- Simple config presets allow training to run on multiple hardware setups with minimal effort.<p>- GRPO used with 16 rollouts per task, up to 32k tokens per rollout.<p>- Agent uses XML&#x2F;YAML format to structure tool calls<p>*More details*:<p>My Github repos open source it all (agent, data, code) and has way more technical details if you are interested!:<p>- Terminal Agent RL repo<p>- Multi-agent synthetic data pipeline repo<p>I thought I would share this because I believe long-horizon RL is going to change everybody&#x27;s lives, and so I feel it is important (and super fun!) for us all to share knowledge around this area, and also have enjoy exploring what is possible.<p>Thanks for reading!<p>Dan<p>(Built using rLLM RL framework which was brilliant to work with, and evaluated and inspired by the great Terminal Bench benchmark)

[Other] Show HN: Rewindtty – Record and replay terminal sessions as structured JSON

[Other] Show HN: Walk-through of rocket landing optimization paper [pdf] Hey all! Long time lurker, first time poster.<p>I found this rocket landing trajectory optimization paper cool, but it took me a while to wrap my head around it and implement it. I wrote up an expanded version of the paper including details that would have helped me understand it the first time through, with the idea being that it might make the content more approachable for others with similar interests. The source code is also linked in the document.<p>I&#x27;m open to feedback, I&#x27;m always trying to get better across the board.

[Other] Show HN: Kiln – AI Boilerplate with Evals, Fine-Tuning, Synthetic Data, and Git I noticed there weren&#x27;t boilerplates for AI projects like there were for web apps, so I built one. Same idea - everything you need to get a project up and running quickly. However, instead of web-framework&#x2F;CSS&#x2F;DB, it&#x27;s tools for AI projects: evals, synthetic data gen, fine-tuning, and more.<p>Kiln is a free, open tool that gives you everything most AI projects need in one integrated package:<p>- Eval system: including LLM-as-judge evals, eval data generation, human baselines<p>- Fine-tuning: proxy to many fine-tuning providers like Fireworks&#x2F;Together&#x2F;OpenAI&#x2F;Unsloth<p>- Synthetic data generation: deeply integrated into evals and fine-tuning<p>- Model routing: 12 providers including Ollama, OpenRouter, and more<p>- Git-based collaboration: projects are designed to be synced through your own git server<p>The key insight is that these tools work much better when they&#x27;re integrated. For example, the synthetic data generator knows whether you&#x27;re creating data for evals vs. fine-tuning (which have very different data needs), and evals can automatically test different prompt&#x2F;model&#x2F;fine-tune combinations.<p>It runs entirely locally - your project data stays in local files, and you control your own git repos. No external services required (though it integrates with them if you want).<p>Main project GitHub: <a href="https:&#x2F;&#x2F;github.com&#x2F;Kiln-AI&#x2F;Kiln">https:&#x2F;&#x2F;github.com&#x2F;Kiln-AI&#x2F;Kiln</a><p>Demo GitHub where I use it to build a &#x27;natural language to ffmpeg command&#x27; demo with evals, fine-tunes, and synthetic data (including demo video): <a href="https:&#x2F;&#x2F;github.com&#x2F;Kiln-AI&#x2F;demos&#x2F;blob&#x2F;main&#x2F;end_to_end_project_demo&#x2F;">https:&#x2F;&#x2F;github.com&#x2F;Kiln-AI&#x2F;demos&#x2F;blob&#x2F;main&#x2F;end_to_end_projec...</a>

[Other] Replacing cron jobs with a centralized task scheduler

[Other] Show HN: I built a free tool to find valuable expired domains using AI Hi HN,<p>I’ve been collecting and analyzing expired domains for years — especially those about to drop. Every day, tens of thousands expire. Most are junk, but a few still have traffic, backlinks, SEO value, or just great names. Finding them used to take hours.<p>Last week I put my internal tools online: <a href="https:&#x2F;&#x2F;pendingdelete.domains" rel="nofollow">https:&#x2F;&#x2F;pendingdelete.domains</a> No login, no paywall Updated daily Combines domain history, traffic, SEO data and AI-driven insights to identify valuable expirations The goal: help spot valuable domains quickly and skip the noise.<p>Still a work-in-progress — would love feedback: Is this useful? What signals or filters would you add? Any UI or speed improvements?<p>Thanks!

[Other] Show HN: I made a tool to generate photomosaics with your pictures Hi HN!<p>I wanted to make some photomosaics for an anniversary gift, but I ended up building this tool and turning it into a website that anyone can use.<p>For those who don’t know, a photomosaic is an image made up of many smaller tile images, arranged in a way that forms a larger, recognisable picture.<p>The best part? Everything runs directly in your browser. No files are uploaded, and there’s no sign-up required.

[Other] ZUSE: IRC terminal client

[Other] Show HN: Open-source physical rack-mounted GUI for home lab I have realized that a lot of people nowadays self-host services and set up home labs with mini racks.<p>One major pain point I have come across personally is to quickly get health status from self-hosted services and machines, and have the ability to headlessly control my Raspberry Pi inside a mini rack.<p>So It got me thinking about building a built-in GUI that users can easily add to their Raspberry Pi nodes in their (mini or full) racks or elsewhere.<p>I have previously designed this GUI for an open source project I have been working on (called Ubo pod: github.com&#x2F;ubopod) and decided to detach&#x2F;decouple the GUI into its own standalone module for this use case.<p>The GUI allows headless control of your Raspberry Pi, monitoring of system resources, and application status.<p>I am designing a new PCB and enclosure as part of this re-design to allow for a new form factor that mounts on server racks.<p>I am recording my journey of re-designing this and I would love to get early feedback from users to better understand what they may need or require from such a solution, specially on the hardware side.<p>The software behind the GUI is quite mature (<a href="https:&#x2F;&#x2F;github.com&#x2F;ubopod&#x2F;ubo_app">https:&#x2F;&#x2F;github.com&#x2F;ubopod&#x2F;ubo_app</a>) and you can actually try it right now without the hardware inside the web browser as shown in the video:<p><a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=9Ob_HDO66_8" rel="nofollow">https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=9Ob_HDO66_8</a><p>All PCB designs are available here:<p><a href="https:&#x2F;&#x2F;github.com&#x2F;ubopod&#x2F;ubo-pcb">https:&#x2F;&#x2F;github.com&#x2F;ubopod&#x2F;ubo-pcb</a>

[Other] Show HN: Flyde 1.0 – Like n8n, but in your codebase Hi HN!<p>I&#x27;m excited to share Flyde 1.0. A big update to the open-source visual programming tool I launched here in March of last year (<a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=39628285">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=39628285</a>).<p>Since Flyde’s launch, there&#x27;s been a huge rise in demand for visual builders, especially for AI-heavy workflows. Visual-programming shines with async and concurrency-heavy logic, which describes most LLM chains perfectly.<p>A few months ago, I tried to capitalize on this trend by launching a commercial version of Flyde called Flowcode (<a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=43830193">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=43830193</a>). It didn&#x27;t go well. I learned the hard way that Flyde’s strength wasn&#x27;t just about flexibility or performance compared to tools like n8n. The real value was always how Flyde fits inside your <i>existing codebase</i>. The launch also helped me understand that there&#x27;s still a big gap: no tool really covers the full lifecycle, from rapid prototyping to deep integration, evaluation, and iteration inside your own projects.<p>So, over the last few months, I worked hard to polish Flyde: - Cleaned up and simplified the nodes API - Made it possible to fork any node for maximum flexibility - Launched a new online playground for quick experimenting and sharing (<a href="https:&#x2F;&#x2F;www.flyde.dev&#x2F;playground" rel="nofollow">https:&#x2F;&#x2F;www.flyde.dev&#x2F;playground</a>) - Created a new CLI tool to speed up development and setup - Fixed a ton of bugs - Simplified the UI&#x2F;UX to make it smoother and less confusing<p>There’s still a lot of missing stuff. Better templates, docs, and nodes, but I think it’s finally stable and useful enough to give it another shot.<p>My plan is to first make sure that Flyde is usable and valuable as an OS project, and then try to provide additional value via “Flyde Studio” - a SaaS that will help non-engineers iterate on Flyde flows from a web-app. Changes become a PR in the host repo.<p>I&#x27;d really love some honest feedback and hear whether Flyde resonates with an existing pain&#x2F;problem.<p>Check it out here: Playground: <a href="https:&#x2F;&#x2F;www.flyde.dev&#x2F;playground" rel="nofollow">https:&#x2F;&#x2F;www.flyde.dev&#x2F;playground</a><p>GitHub: <a href="https:&#x2F;&#x2F;github.com&#x2F;flydelabs&#x2F;flyde">https:&#x2F;&#x2F;github.com&#x2F;flydelabs&#x2F;flyde</a><p>Looking forward to hearing your thoughts! - Gabriel

[Code Quality] Show HN: Dlg – Zero-cost printf-style debugging for Go Hey HN,<p>I tend to use printf-style debugging as my primary troubleshooting method and only resort to gdb as a last resort.<p>While I like its ease of use printf debugging isn&#x27;t without its annoyances, namely removing the print statements once you&#x27;re done.<p>I used to use trace-level logging from proper logging libraries but adding trace calls in every corner quickly gets out of control and results in an overwhelming amount of output.<p>To scratch my own itch I created dlg - a minimal debugging library that disappears completely from production builds. Its API exposes just a single function, Printf [1].<p>dlg is optimized for performance in debug builds and, most importantly, when compiled without the dlg build tag, all calls are eliminated by the Go linker as if dlg was never imported.<p>For debug builds it adds optional stack trace generation configurable via environment variables or linker flags.<p>GitHub: <a href="https:&#x2F;&#x2F;github.com&#x2F;vvvvv&#x2F;dlg">https:&#x2F;&#x2F;github.com&#x2F;vvvvv&#x2F;dlg</a><p>Any feedback is much appreciated.<p>[1]: Actually two functions - there&#x27;s also SetOutput.

[Other] Performance and Telemetry Analysis of Trae IDE, ByteDance's VSCode Fork Hi HN, I was evaluating IDEs for a personal project and decided to test Trae, ByteDance&#x27;s fork of VSCode. I immediately noticed some significant performance and privacy issues that I felt were worth sharing. I&#x27;ve written up a full analysis with screenshots, network logs, and data payloads in the linked post.<p>Here are the key findings:<p>1. Extreme Resource Consumption: Out of the box, Trae used 6.3x more RAM (~5.7 GB) and spawned 3.7x more processes (33 total) than a standard VSCode setup with the same project open. The team has since made improvements, but it&#x27;s still significantly heavier.<p>2. Telemetry Opt-Out Doesn&#x27;t Work (It Makes It Worse): I found Trae was constantly sending data to ByteDance servers (byteoversea.com). I went into the settings and disabled all telemetry. To my surprise, this didn&#x27;t stop the traffic. In fact, it increased the frequency of batch data collection. The telemetry &quot;off&quot; switch appears to be purely cosmetic.<p>3. What&#x27;s Being Sent: Even with telemetry &quot;disabled,&quot; Trae sends detailed payloads including: Hardware specs (CPU, memory, etc.) Persistent user, device, and machine IDs OS version, app language, user name Granular usage data like time-on-ide, window focus state, and active file types.<p>4. Community Censorship: When I tried to discuss these findings on their official Discord, my posts were deleted and my account was muted for 7 days. It seems words like &quot;track&quot; trigger an automated gag rule, which prevents any real discussion about privacy.<p>I believe developers should be aware of this behavior. The combination of resource drain, non-functional privacy settings, and censorship of technical feedback is a major red flag. The full, detailed analysis with all the evidence (process lists, Fiddler captures, JSON payloads, and screenshots of the Discord moderation) is available at the link. Happy to answer any questions.

[Other] Show HN: Cant, rust nn lib for learning Hey! This is something i have been working on. A tiny neural networking lib to learn how something like pytorch works, and to improve my own coding standards.