Hey HN, Waleed here. We&#x27;re building Sim (<a href="https:&#x2F;&#x2F;sim.ai&#x2F;">https:&#x2F;&#x2F;sim.ai&#x2F;</a>), an open-source visual editor to build agentic workflows. Repo here: <a href="https:&#x2F;&#x2F;github.com&#x2F;simstudioai&#x2F;sim&#x2F;" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;simstudioai&#x2F;sim&#x2F;</a>. Docs here: <a href="https:&#x2F;&#x2F;docs.sim.ai">https:&#x2F;&#x2F;docs.sim.ai</a>.You can run Sim locally using Docker, with no execution limits or other restrictions.We started building Sim almost a year ago after repeatedly troubleshooting why our agents failed in production. Code-first frameworks felt hard to debug because of implicit control flow, and workflow platforms added more overhead than they removed. We wanted granular control and easy observability without piecing everything together ourselves.We launched Sim [1][2] as a drag-and-drop canvas around 6 months ago. Since then, we&#x27;ve added:- 138 blocks: Slack, GitHub, Linear, Notion, Supabase, SSH, TTS, SFTP, MongoDB, S3, Pinecone, ...- Tool calling with granular control: forced, auto- Agent memory: conversation memory with sliding window support (by last n messages or tokens)- Trace spans: detailed logging and observability for nested workflows and tool calling- Native RAG: upload documents, we chunk, embed with pgvector, and expose vector search to agents- Workflow deployment versioning with rollbacks- MCP support, Human-in-the-loop block- Copilot to build workflows using natural language (just shipped a new version that also acts as a superagent and can call into any of your connected services directly, not just build workflows)Under the hood, the workflow is a DAG with concurrent execution by default. Nodes run as soon as their dependencies (upstream blocks) are satisfied. Loops (for, forEach, while, do-while) and parallel fan-out&#x2F;join are also first-class primitives.Agent blocks are pass-through to the provider. You pick your model (OpenAI, Anthropic, Gemini, Ollama, vLLM), and and we pass through prompts, tools, and response format directly to the provider API. We normalize response shapes for block interoperability, but we&#x27;re not adding layers that obscure what&#x27;s happening.We&#x27;re currently working on our own MCP server and the ability to deploy workflows as MCP servers. Would love to hear your thoughts and where we should take it next :)[1] <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=43823096">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=43823096</a>[2] <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=44052766">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=44052766</a>

Show HN: Sim – Apache-2.0 n8n alternative

Hi HN!I&#x27;m building a JavaScript debugger called Wirebrowser. It combines network inspection, request rewriting, heap snapshots, and live object search.The main experimental feature is BDHS (Breakpoint-Driven Heap Search): it hooks into the JavaScript debugger and automatically captures a heap snapshot at every pause and performs a targeted search for the value or structure of interest.
This reveals the moment a value appears in memory and the user-land function responsible for creating it.Another interesting feature is the Live Object Search: it inspects runtime objects (not just snapshots), supports regex and object similarity, and lets you patch objects directly at runtime.Whitepaper:
<a href="https:&#x2F;&#x2F;fcavallarin.github.io&#x2F;wirebrowser&#x2F;BDHS-Origin-Trace" rel="nofollow">https:&#x2F;&#x2F;fcavallarin.github.io&#x2F;wirebrowser&#x2F;BDHS-Origin-Trace</a>Feedback very welcome, especially on whether BDHS would help your debugging workflow.

Show HN: Wirebrowser – A JavaScript debugger with breakpoint-driven heap search

Mobile keyboards today are almost entirely based on the 26-key, 3-row QWERTY layout.
Here’s a new 2-row, 16-key alternative designed specifically for smartphones.

Show HN: A 2-row, 16-key keyboard designed for smartphones

I built Read Fast after realising I was saving hundreds of things to read and reading almost none of them.It automatically turns saved content into a scheduled briefing and sends it to my inbox at whatever time I choose.New version since last post actually allows you to bulk upload URLs. So you can just dump links here and Read Fast will scrape and brief for you.Great for actually reading the things I want to. Curious if that resonates with people?

Show HN: I built time to read all the things I want to

Hey HN! Like most here regular meetings have always been a big part of my work.Over the years I&#x27;ve learned the value of active note taking in these meetings. Meaning: not minutes, not transcriptions or AI summaries, but me using my brain to actively pull out the key points in short form bullet-like notes, as the meeting is going on, as I&#x27;m talking and listening (and probably typing with one hand). This could be agenda points to cover, any interesting sidebars raised, insights gotten to in a discussion, actions agreed to (and a way to track whether they got done next time!).It&#x27;s both useful just to track what&#x27;s going on in all these different meetings week to week (at one point I was doing about a dozen 1-1s per week, and it just becomes impossible to hold it in RAM) but also really valuable over time when you can look back and see the full history of a particular meeting, what was discussed when, how themes and structure are changing, is the meetings effective, etc.Anyway, I&#x27;ve tried a bunch of different tools for taking these notes over the years. All the obvious ones you&#x27;ve probably used too. And I&#x27;ve always just been not quite satisfied with the experience. They work, obviously (it&#x27;s just text based notes at the end of the day) but nothing is first-class for this usecase.So, I decided to build the tool I&#x27;ve always felt I want to use, specifically for regular 1-1s and other types of regular meetings. I&#x27;ve been using it myself and with friends for a while already now, and I think it&#x27;s got to that point where I actually prefer to reach for it over other general purpose note taking tools now, and I want to share it more widely.There&#x27;s a free tier so you can use it right away, in fact without even signing up.If you&#x27;ve also been wanting a better system to manage your notes for regular meetings, give it a go and let me know what you think!

Show HN: I built a system for active note-taking in regular meetings like 1-1s

I built AlgoDrill because I kept grinding LeetCode, thinking I knew the pattern, and then completely blanking when I had to implement it from scratch a few weeks later.AlgoDrill turns NeetCode 150 and more into pattern-based drills: you rebuild the solution line by line with active recall, get first principles editorials that explain why each step exists, and everything is tagged by patterns like sliding window, two pointers, and DP so you can hammer the ones you keep forgetting. The goal is simple: turn familiar patterns into code you can write quickly and confidently in a real interview.<a href="https:&#x2F;&#x2F;algodrill.io" rel="nofollow">https:&#x2F;&#x2F;algodrill.io</a>Would love feedback on whether this drill-style approach feels like a real upgrade over just solving problems once, and what’s most confusing or missing when you first land on the site.

Show HN: AlgoDrill – Interactive drills to stop forgetting LeetCode patterns

Built this over the last few days, based on a Rust codebase that parses the latest ALPR reports from OpenStreetMaps, calculates navigation statistics from every tagged residential building to nearby amenities, and tests each route for intersection with those ALPR cameras (Flock being the most widespread).These have gotten more controversial in recent months, due to their indiscriminate large scale data collection, with 404 Media publishing many original pieces (<a href="https:&#x2F;&#x2F;www.404media.co&#x2F;tag&#x2F;flock&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.404media.co&#x2F;tag&#x2F;flock&#x2F;</a>) about their adoption and (ab)use across the country. I wanted to use open source datasets to track the rapid expansion, especially per-county, as this data can be crucial for &#x27;deflock&#x27; movements to petition counties and city governments to ban and remove them.In some counties, the tracking becomes so widespread that most people can&#x27;t go anywhere without being photographed. This includes possibly sensitive areas, like places of worship and medical facilities.The argument for their legality rests upon the notion that these cameras are equivalent to &#x27;mere observation&#x27;, but the enormous scope and data sharing agreements in place to share and access millions of records without warrants blurs the lines of the fourth amendment.

Show HN: Automated license plate reader coverage in the USA

Hi there, HN! We’re Jai and Sanket from DeepSource (YC W20), and today we’re launching Autofix Bot, a hybrid static analysis + AI agent purpose-built for in-the-loop use with AI coding agents.AI coding agents have made code generation nearly free, and they’ve shifted the bottleneck to code review. Static-only analysis with a fixed set of checkers isn’t enough. LLM-only review has several limitations: non-deterministic across runs, low recall on security issues, expensive at scale, and a tendency to get ‘distracted’.We spent the last 6 years building a deterministic, static-analysis-only code review product. Earlier this year, we started thinking about this problem from the ground up and realized that static analysis solves key blind spots of LLM-only reviews. Over the past six months, we built a new ‘hybrid’ agent loop that uses static analysis and frontier AI agents together to outperform both static-only and LLM-only tools in finding and fixing code quality and security issues. Today, we’re opening it up publicly.Here’s how the hybrid architecture works:- Static pass: 5,000+ deterministic checkers (code quality, security, performance) establish a high-precision baseline. A sub-agent suppresses context-specific false positives.- AI review: The agent reviews code with static findings as anchors. Has access to AST, data-flow graphs, control-flow, import graphs as tools, not just grep and usual shell commands.- Remediation: Sub-agents generate fixes. Static harness validates all edits before emitting a clean git patch.Static solves key LLM problems: non-determinism across runs, low recall on security issues (LLMs get distracted by style), and cost (static narrowing reduces prompt size and tool calls).On the OpenSSF CVE Benchmark [1] (200+ real JS&#x2F;TS vulnerabilities), we hit 81.2% accuracy and 80.0% F1; vs Cursor Bugbot (74.5% accuracy, 77.42% F1), Claude Code (71.5% accuracy, 62.99% F1), CodeRabbit (59.4% accuracy, 36.19% F1), and Semgrep CE (56.9% accuracy, 38.26% F1). 
On secrets detection, 92.8% F1; vs Gitleaks (75.6%), detect-secrets (64.1%), and TruffleHog (41.2%). We use our open-source classification model for this. [2]Full methodology and how we evaluated each tool: <a href="https:&#x2F;&#x2F;autofix.bot&#x2F;benchmarks" rel="nofollow">https:&#x2F;&#x2F;autofix.bot&#x2F;benchmarks</a>You can use Autofix Bot interactively on any repository using our TUI, as a plugin in Claude Code, or with our MCP on any compatible AI client (like OpenAI Codex).[3] We’re specifically building for AI coding agent-first workflows, so you can ask your agent to run Autofix Bot on every checkpoint autonomously.Give us a shot today: <a href="https:&#x2F;&#x2F;autofix.bot" rel="nofollow">https:&#x2F;&#x2F;autofix.bot</a>. We’d love to hear any feedback!---[1] <a href="https:&#x2F;&#x2F;github.com&#x2F;ossf-cve-benchmark&#x2F;ossf-cve-benchmark" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;ossf-cve-benchmark&#x2F;ossf-cve-benchmark</a>[2] <a href="https:&#x2F;&#x2F;huggingface.co&#x2F;deepsource&#x2F;Narada-3.2-3B-v1" rel="nofollow">https:&#x2F;&#x2F;huggingface.co&#x2F;deepsource&#x2F;Narada-3.2-3B-v1</a>[3] <a href="https:&#x2F;&#x2F;autofix.bot&#x2F;manual&#x2F;#terminal-ui" rel="nofollow">https:&#x2F;&#x2F;autofix.bot&#x2F;manual&#x2F;#terminal-ui</a>

Show HN: Autofix Bot – Hybrid static analysis and AI code review agent

Hi HN, I&#x27;ve always been interested in fractals, especially the Mandelbrot and Julia sets. A few years ago I created a 2d viewer of this inherently 4d space. But the other day I decided to ask Claude and GPT how to make this a full RT 3d explorer. A few hours later and this was vibe coded.To use it you can use the mouse to rotate the fractal the the mouse wheel to zoom in and out. to map from 4d to 3d, one of the dims is mapped to an adjustable slider. the there is also a clipping plane slider to help visualize the internal structures of the fractal.I have mixed feelings about vibe coding. It was amazing to go from an idea to live implementation within a few hours, but in my coding projects, I&#x27;ve always appreciated the journey and the learning, not just the final product. Vibe coding kind of skips to the end which is exciting and efficient, but just not as fulfilling as struggling through a project step-by-step.

Show HN: A real-time 4D fractal explorer in the browser using WebGPU

Show HN: KV and wide-column database with CDN-scale replication

Show HN: I built this after leaking my AWS keys on StackOverflow

Show HN: I built a one-click coin flip with no ads or tracking

I&#x27;m building an open source Amazon.In other words, an open source decentralized marketplace. But like Carl Sagan said, to make an apple pie from scratch, you must first invent the universe.So first I had to make open source management systems for every vertical. I&#x27;m launching the first one today, Openfront e-commerce, an open source Shopify alternative. Next will be Openfront restaurant, Openfront grocery, and Openfront gym.And all of these Openfronts will connect to our decentralized marketplace, &quot;the&#x2F;marketplace&quot;, seamlessly. Once we launch other Openfronts, you&#x27;ll be able to do everything from booking hotels to ordering groceries right from one place with no middle men. The marketplace simply connects to the Openfront just like its built-in storefront does.Together, we can use open source to disrupt marketplaces and make sure sellers, in every vertical, are never beholden to them.Marketplace: <a href="https:&#x2F;&#x2F;marketplace.openship.org" rel="nofollow">https:&#x2F;&#x2F;marketplace.openship.org</a>Openfront platforms: <a href="https:&#x2F;&#x2F;openship.org&#x2F;openfront-ecommerce" rel="nofollow">https:&#x2F;&#x2F;openship.org&#x2F;openfront-ecommerce</a>Source code: <a href="https:&#x2F;&#x2F;github.com&#x2F;openshiporg&#x2F;openfront" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;openshiporg&#x2F;openfront</a>Demo - Openfront: <a href="https:&#x2F;&#x2F;youtu.be&#x2F;jz0ZZmtBHgo" rel="nofollow">https:&#x2F;&#x2F;youtu.be&#x2F;jz0ZZmtBHgo</a>Demo - Marketplace: <a href="https:&#x2F;&#x2F;youtu.be&#x2F;LM6hRjZIDcs" rel="nofollow">https:&#x2F;&#x2F;youtu.be&#x2F;LM6hRjZIDcs</a>Part 1 - <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=32690410">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=32690410</a>

Show HN: I'm building an open-source Amazon

The biggest problem with Tic-Tac-Toe is that it almost always ends in a draw. Tic Tac Flip tries to fix that!Learn the rules in Learning Mode or below:- Winning Criteria: 3 Ghosts (Flipped O or X, which can be a mixture). It&#x27;s not just 3 Os or 3 Xs anymore!- Flipping Mechanic: When one or more lines having only O and X are formed, the minority of either all Os or all Xs get flipped to a Ghost, and the majority gets removed from the board. E.g., A line of 2 Os and 1 X leads to 1 X ghost and the removal of 2 Os.- Active Flip: You can actively flip your O&#x2F;X to a Ghost (or flip a ghost back) once per game.- Placing Ghost Directly: You can place a &quot;Ghost&quot; piece directly as a final winning move (only once, and only when there are two existing ghosts in a line).I&#x27;m looking for feedback on the game balance and learning curve. Specifically:
- Is the &quot;Ghost&quot; and &quot;Flip&quot; mechanic intuitive?
- Is the Learning Mode helpful?
- Is the game fair? Any rule adjustments needed?
- Any bugs or issues?Any suggestions or comments would be much appreciated. Thank you in advance!

Show HN: Tic Tac Flip – A new strategic game based on Tic Tac Toe

wget <a href="https:&#x2F;&#x2F;github.com&#x2F;beehive-lab&#x2F;TornadoVM&#x2F;releases&#x2F;download&#x2F;v2.1.0&#x2F;tornadovm-2.1.0-opencl-linux-amd64.zip" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;beehive-lab&#x2F;TornadoVM&#x2F;releases&#x2F;download&#x2F;v...</a>
unzip tornadovm-2.1.0-opencl-linux-amd64.zip
# Replace &lt;path-to-sdk&gt; manually with the absolute path of the extracted folder
export TORNADO_SDK=&quot;&lt;path-to-sdk&gt;&#x2F;tornadovm-2.1.0-opencl&quot;
export PATH=$TORNADO_SDK&#x2F;bin:$PATHtornado --devices
tornado --version# Navigate to the project directory
cd GPULlama3.java# Source the project-specific environment paths -&gt; this will ensure the 
source set_paths# Build the project using Maven (skip tests for faster build)
# mvn clean package -DskipTests or just make
make# Run the model (make sure you have downloaded the model file first - see below)
.&#x2F;llama-tornado --gpu --verbose-init --opencl --model beehive-llama-3.2-1b-instruct-fp16.gguf --prompt &quot;tell me a joke&quot;

Show HN: GPULlama3.java Llama Compilied to PTX/OpenCL Now Integrated in Quarkus

I built Jottings because I was tired of my own thoughts getting trapped inside algorithmic feeds where I had to perform. There was a huge mental load before posting something on X or Instagram.Every time I wanted to share something small or unfinished, I opened Twitter and lost 20 minutes to the timeline. Writing a blog post felt too heavy for those smaller, quick thoughts. I just wanted a place to write something down quickly and hit publish.Jottings is that place. It gives you a clean microblog on a domain you own. Posts show up in simple chronological order. No likes. No followers. No feed trying to decide what matters.What Jottings is
- A microblogging platform that builds fully static microblog sites
- A free subdomain (you.jottings.me) or connect your own domain on PRO plans
- Markdown, tags, RSS feed, links with preview, and image uploads
- An optional AI writing helper when you are stuck or lazy to fix grammar
- Optimized for SEO and AI search friendly
- Analytics for your sitesWhat it is not
- Not a social network
- Not an engagement funnel
- Not trying to keep you on the site
- Not a replacement for long-form blogging, though you can use it that wayHow it works
Each Jot publish triggers a static site rebuild. The site is stored in Cloudflare R2 and served directly at the edge. Custom domains go through Cloudflare SSL for SaaS.
I built it to be boring, reliable (barring Cloudflare&#x27;s latest issues), and cheap to run.Pricing
Free tier for a subdomain, text posts, and a lot more.
USD5 per month for custom domains, images, full Markdown, and the writing helper.
I priced it to be an easy yes.Limitations
- No comments (on purpose)
- No native apps yet (iOS is coming)
- The writing helper is helpful but not magic
- I am a solo founder, so features move at human speedI use Jottings regularly to document what I build. It has been the lowest-friction way I have found to publish anything publicly.Demo of Jottings site for product updates: <a href="https:&#x2F;&#x2F;jottings.jottings.me&#x2F;" rel="nofollow">https:&#x2F;&#x2F;jottings.jottings.me&#x2F;</a>
Demo of my personal Jottings site: <a href="https:&#x2F;&#x2F;jottings.vishalvshekkar.com" rel="nofollow">https:&#x2F;&#x2F;jottings.vishalvshekkar.com</a> (with custom subdomain)I would love feedback from this community. What would make this better or more useful for you?Check it out here: <a href="https:&#x2F;&#x2F;jottings.me" rel="nofollow">https:&#x2F;&#x2F;jottings.me</a> (2 min set up)
Feel free to write to me at vishal@vishalvshekkar.com— Vishal

Show HN: Jottings; Anti-social microblog for your thoughts

I’ve been working on gotui, a modern fork of the unmaintained termui, rebuilt on top of tcell for TrueColor, mouse support, and proper resize handling. It keeps the simple termui-style API, but adds a bunch of new widgets (charts, gauges, world map, etc.), nicer visuals (collapsed borders, rounded corners), and input components for building real dashboards and tools. Under the hood the renderer’s been reworked for much better performance, and I’d love feedback on what’s missing for you to use it in production.

Show HN: Gotui – a modern Go terminal dashboard library

I built a procedurally generated word-search game where the puzzle never ends - as you scroll, the grid expands infinitely and new words appear. It’s designed to be quick to pick up, satisfying to play, and a little addictive.The core game works without an account using the pre-defined games, but signing up allows you to generate games using any topic you can think of.I’d love feedback on gameplay, performance, and whether the endless format feels engaging over time.
If you try it, I’d really appreciate any bug reports or suggestions.Thanks in advance!

Show HN: An endless scrolling word search game

Powerlifting programming isn’t as complicated as spreadsheets make it look; once you strip away the noise, the structure fits comfortably on a phone. I think of it as two parts: plain sets&#x2F;weight&#x2F;reps&#x2F;RPE&#x2F;1RM, and the custom analytics you build on top. SigmaLifting handles the first part entirely on your phone with a mobile-first UI and a flexible data model that can express most spreadsheet-style programs. For the second part, you export the data and run it through your own tools (or an LLM) for whatever analysis you want.

Show HN: SigmaLifting – A protocol for powerlifting training data

Hello Everyone! We built SQLFlow as a lightweight stream processing engine.We leverage DuckDB as the stream processing engine, which gives SQLFlow the ability to process 10&#x27;s of thousands of messages a second using ~250MiB of memory!DuckDB also supports a rich ecosystem of sinks and connectors!<a href="https:&#x2F;&#x2F;sql-flow.com&#x2F;docs&#x2F;category&#x2F;tutorials&#x2F;" rel="nofollow">https:&#x2F;&#x2F;sql-flow.com&#x2F;docs&#x2F;category&#x2F;tutorials&#x2F;</a><a href="https:&#x2F;&#x2F;github.com&#x2F;turbolytics&#x2F;sql-flow" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;turbolytics&#x2F;sql-flow</a>We were tired of running JVM&#x27;s for simple stream processing, and also of bespoke one off stream processorsI would love your feedback, criticisms and&#x2F;or experiences!Thank you

Show HN: DuckDB for Kafka Stream Processing

Hi!I got tired of setting up tools I can&#x27;t explain to a team in a few words like sops or git-crypt, just to store few files with environment variables or secrets, so I built lockenv as a simple alternative.It&#x27;s basically a password-protected vault file you commit to git. No gpg keys, no cloud, just lockenv init, set a password, and lock&#x2F;unlock the secrets.This tool integrates with OS keyring, so you&#x27;re not typing passwords constantly. Should work on Mac&#x2F;Linux&#x2F;Windows, but I tested it only on linux so far.I am not trying to replace any mature &#x2F; robust solution, just making small tool for simple cases, where I want to stop sharing secrets via slack.Feel free to try, thank you!

Show HN: Lockenv – Simple encrypted secrets storage for Git

Show HN: Soup.lua: making Lua do what it shouldn't

My kids are surprisingly keen to learn to type, and I couldn&#x27;t find any thing out there which wasA) Free without adverts
B) Worked well on mobile&#x2F;tablets
C) Was clutter free and easy to useI wondered how easily I could vibe code a solution.Here&#x27;s the result.Everything here was &quot;vibe coded&quot; to an extent, the graphics, sounds, art-work, even the github pipelines ( and I used chatgpt to instruct me how to configure everything in AWS )I&#x27;m particularly pleased with the phoneme game, <a href="https:&#x2F;&#x2F;free-kids-typing-games.com&#x2F;games&#x2F;phoneme-sound-lab&#x2F;" rel="nofollow">https:&#x2F;&#x2F;free-kids-typing-games.com&#x2F;games&#x2F;phoneme-sound-lab&#x2F;</a> which was crafted using Google&#x27;s TTS engine, and I think could be expanded further.The mobile keyboard is also particularly impressive and works really well for little fingers.I thought the hacker news crowd might be interested.You can see the code in all it&#x27;s gory AI generated detail here<a href="https:&#x2F;&#x2F;github.com&#x2F;Alan01252&#x2F;free-kids-typing-games.com" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;Alan01252&#x2F;free-kids-typing-games.com</a>

Show HN: I vibe coded a free typing game for my kids

My daily driver is a Macbook Air M2 running Linux - Fedora Asahi Remix to be precise. One thing I missed when using it is the ability to occasionally run Windows software using Wine. Apparently, you can run Steam on it and apparently Steam allows installing and running arbitrary Windows software, but when I tried it, I couldn&#x27;t create an account and in general, I&#x27;d rather not use Stream. I succeeded running and older version of Heroic Games Launcher [1] under muvm (a virtual machine that runs a 4K page kernel on a 16K one). That wasn&#x27;t terribly easy though and I wanted a better experience. My other problem with Heroic and with other launchers focused on games is the lack of flexibility - they either work for a particular piece of software or they don&#x27;t, and you can&#x27;t do anything about it. For instance, an installer may require a certain package to be installed using Winetricks [2], before it runs. Heroic doesn&#x27;t give you a chance to run anything before the installer runs. Long story short, I decided to build my own Wine prefix manager that would be flexible, not focused exclusively on games and would run on Asahi Linux. Also, I decided to write it in a new language for me (Dart &#x2F; Flutter) and learn that language as a byproduct. Five months later, it&#x27;s finally ready and I&#x27;d like to get feedback on it. BTW, it also supports regular x86_64 Linux distros, though it doesn&#x27;t receive as much testing on them.[1]: <a href="https:&#x2F;&#x2F;heroicgameslauncher.com&#x2F;" rel="nofollow">https:&#x2F;&#x2F;heroicgameslauncher.com&#x2F;</a>
[2]: <a href="https:&#x2F;&#x2F;github.com&#x2F;Winetricks&#x2F;winetricks" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;Winetricks&#x2F;winetricks</a>

Show HN: WineBar: A yet another Wine prefix manager, with Asahi Linux support

I built this to explore low-latency C++ patterns.It uses a sharded architecture (one thread per symbol group) to avoid mutex contention during matching. Recently moved to int64_t fixed-point pricing and added memory compaction to prevent leaks in long-running processes.

Show HN: High-Performance Order Matching Engine in C++20 (2.2M ops/SEC)

This project reconstructs the Epstein email records from the recent U.S. House Oversight Committee releases using only public-domain documents (23,124 image files + 2,800 OCR text files).Most email pages contain only one real message, buried under layers of repeated headers&#x2F;footers. I wanted to rebuild the conversations without all the surrounding noise.I used an OCR + vision-LLM pipeline to extract individual messages from the email screenshots, normalize senders&#x2F;recipients, rebuild timestamps, detect duplicates, and map threads. The output is a structured SQLite database that runs client-side via SQL.js (WebAssembly).The repository includes the full extraction pipeline, data cleaning scripts, schema, limitations, and implementation notes. The interface is a lightweight PWA that displays the reconstructed messages in a phone-style UI, with links back to every original source image for verification.Live demo: <a href="https:&#x2F;&#x2F;epsteinsphone.org" rel="nofollow">https:&#x2F;&#x2F;epsteinsphone.org</a>All source data is from the official public releases; no leaks or private material.Happy to answer questions about the pipeline, LLM extraction, threading logic, or the PWA implementation.

Show HN: Epstein's emails reconstructed in a message-style UI (OCR and LLMs)

Hi HN, tl;dr we built a bug finder that&#x27;s working really well, especially for app backends. Try it out and send us your thoughts!Long story below.--------------------------We originally set out to work on technical debt. We had all seen codebases with a lot of debt, so we had personal grudges about the problem, and AI seemed to be making it a lot worse.Tech debt also seemed like a great problem for AI because: 1) a small portion of the work is thinky and strategic, and then the bulk of the execution is pretty mechanical, and 2) when you&#x27;re solving technical debt, you&#x27;re usually trying to preserve existing behavior, just change the implementation. That means you can treat it as a closed-loop problem if you figure out good ways to detect unintended behavior changes due to a code change. And we know how to do that – that&#x27;s what tests are for!So we started with writing tests. Tests create the guardrails that make future code changes safer. Our thinking was: if we can test well enough, we can automate a lot of other tech debt work at very high quality.We built an agent that could write thousands of new tests for a typical codebase, most &quot;merge-quality&quot;. Some early users merged hundreds of PRs generated this way, but intuitively the tool always felt &quot;good but not great&quot;. We used it sporadically ourselves, and it usually felt like a chore.Around this point we realized: while we had set out to write good tests, we had built a system that, with a few tweaks, might be very good at finding bugs. When we tested it out on some friends&#x27; codebases, we discovered that almost every repo has tons of bugs lurking in it that we were able to flag. Serious bugs, interesting enough that people dropped what they were doing to fix them. Sitting right there in peoples codebases, already merged, running in prod.We also found a lot of vulns, even in mature codebases, and sometimes even right after someone had gotten a pentest.Under the hood:
- We check out a codebase and figure out how to build it for local dev and exercise it with tests.
- We take snapshots of the built local dev state. (We use Runloop for this and are big fans.)
- We spin up hundreds of copies of the local dev environment to exercise the codebase in thousands of ways and flag behaviors that seem wrong.
- We pick the most salient, scary examples and deliver them as linear tickets, github issues, or emails.In practice, it&#x27;s working pretty well. We&#x27;ve been able to find bugs in everything from compilers to trading platforms (even in rust code), but the sweet spot is app backends.Our approach trades compute for quality. Our codebase scans take hours, far beyond what would be practical for a code review bot. But the result is that we can make more judicious use of engineers’ attention, and we think that’s going to be the most important variable.Longer term, we think compute is cheap, engineer attention is expensive. Wielded properly, the newest models can execute complicated changes, even in large codebases. That means the limiting reagent in building software is human attention. It still takes time and focus for an engineer to ingest information, e.g. existing code, organizational context, and product requirements. These are all necessary before an engineer can articulate what they want in precise terms and do a competent job reviewing the resulting diff.For now we&#x27;re finding bugs, but the techniques we&#x27;re developing extend to a lot of other background, semi-proactive work to improve codebases.Try it out and tell us what you think. Free first scan, no credit card required: <a href="https:&#x2F;&#x2F;detail.dev&#x2F;" rel="nofollow">https:&#x2F;&#x2F;detail.dev&#x2F;</a>We&#x27;re also scanning on OSS repos, if you have any requests. The system is pretty high signal-to-noise, but we don&#x27;t want to risk annoying maintainers by automatically opening issues, so if you request a scan for an OSS repo the results will go to you personally. <a href="https:&#x2F;&#x2F;detail.dev&#x2F;oss" rel="nofollow">https:&#x2F;&#x2F;detail.dev&#x2F;oss</a>

Show HN: Detail, a Bug Finder

<a href="https:&#x2F;&#x2F;github.com&#x2F;mcp-shark&#x2F;mcp-shark" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;mcp-shark&#x2F;mcp-shark</a>
Site: <a href="https:&#x2F;&#x2F;mcpshark.sh&#x2F;" rel="nofollow">https:&#x2F;&#x2F;mcpshark.sh&#x2F;</a>I built MCPShark, a traffic inspector for the Model Context Protocol (MCP).It sits between your editor&#x2F;LLM client and MCP servers so you can:
 • See all MCP traffic (requests, responses, tools, resources) in one place
 • Debug sessions when tools don’t behave as expected
 • Optionally run “Smart Scan” checks to flag risky tools &#x2F; configs