Next I’m working on making reviewing large AI-generated PRs easier, but haven’t gotten there yet.
It’s coming along great. I’ve only been working on it since February, and it’s very complete, functionally. I’ve been using an LLM, extensively.
The original server took over seven months to write, but with the LLM, I had it done in about a month. The PHP code is great quality.
It’s not all bunnies and flowers, though. I’m spending the next few days, rewriting a Swift viewcontroller that the LLM wrote. It works, but the code is awful. It looks like it was written by a jargon-addled high schooler on Adderall. I just can’t bear to have that level of junk in my app. I asked it to refactor the code to improve the quality, and reduce the size, and it removed all the documentation and logging; leaving the thread- and delay-riddled garbage behind.
If someone that worked for me had done that, I would have evicted them, with a trebuchet.
I realized that I can't count on the LLM to help me maintain its mess, so I'd better just suck it up, and do it myself.
The good news is, is that I’m so far ahead of schedule (a position that I’ve never been in, in forty years of development), that I can afford to do this.
- spec driven development workflow: https://github.com/sermakarevich/sddw
- fleet, orchestrator for running many coding agents: https://github.com/sermakarevich/fleet
- chunker, which builds smart chunks of the documents with hierarchical organization: https://github.com/sermakarevich/chunker
- ai knowledge wiki, a collection of ai papers and articles organized in hierarchical Wiki
- hierarchical organization of all passed Kaggle competition solutions (write ups and notebooks) - https://github.com/sermakarevich/kaggle_wiki
- claude code workflows, plugin to build custom claude code workflows based on ssd idea: https://github.com/sermakarevich/ccw
I sometimes need to have a quick but realistic model of an optical system without paying a few thousand for some of the well known commercial offerings, so I've been building this.
To do that, I'm building sharc, a port of Hacker News that runs on Common Lisp, implementing all the latest features of HN. (That last bit is the hard part.) https://github.com/shawwn/sharc
I just implemented collapsing comments and root/next/prev/context nav links. Also (sitename "https://x.com/jsrailton") now returns "x.com/jsrailton" instead of "x.com". (About to implement HN's "from" endpoint, e.g. https://news.ycombinator.com/from?site=twitter.com/jsrailton)
You can read the entire changelog here: https://github.com/shawwn/sharc/tree/main/docs/agents/handof...
In theory, it's really not suited for this because (at least back then) this was designed for computers with:
- relatively large amounts of RAM (uncompressed color+heightmap data takes a lot of space),
- fast CPUs with data caches (it's a 2D trapezoidal walk of a top-down projected fustrum with plenty of additions, multiplications and divisions),
- memory-mapped linear framebuffers (for software rendering).
The PlayStation has none of these things: only 2 MiB of main RAM; an in-order scalar 33 MHz MIPS III processor without a FPU or a data cache, where any CPU load from main RAM stalls the pipeline for 5 cycles, multiplications have a 6 to 13 cycles latency and divisions have a 36 cycles latency; and the only way to touch VRAM is to send commands to the 2D GPU.
What it does have is 4 KiB of I-cache (direct-mapped), 1 KiB of scratchpad with no wait states and a fixed-point GTE coprocessor which is mostly geared towards projecting 3D points onto a 2D screen and not general matrix/vector processing. Meaning that if I get my hot loop within 1024 instructions (and no function calls), fit my working set+stack within 1 KiB and can pipeline the GTE for transformations and the CPU for map scanning/GPU submissions, it might just work.
So far I'm getting decent framerates (like 10-15 FPS at 256x240) with just software projection and GPU line rendering. Right now I'm wrestling with the GTE and various ways to trick it into transforming more points than it's supposed to be able to do, by looking at the fixed-hardware math equations for the various instructions and trying to fit my equations into them (doing stuff like putting relative altitude into X/Y vector registers instead of coordinates and massaging the rest of the values to get screen Y coordinates out of it).
Kind of a hardcore topic to pick for my first homebrew ever, but I craved a simpler, low-level optimization challenge as a pipe cleaner, after working on ghidra-delinker-extension for so long.
A tiny RISC-V linux-based (buildroot) portable audio player hardware with Software written in Rust / Slint.
Inspired by the iPod Nano 7g and totally WIP.
Will probably never be finished but it is a neat learning project
A user reports a location bug on their morning commute. You're 2,000 miles away. Instead of flying out to reproduce it, you replay their exact GPS trip on your simulator. Same route. Same speed. Same timing.
Record & replay real GPS routes for iOS.
Video here: https://x.com/LyleMakes/status/2064438713124270233
VT Code: https://github.com/vinhnx/VTCode > An open-source coding agent with LLM-native code understanding and robust shell safety. Supports multiple LLM providers with automatic failover and efficient context management. => Recently I got 500th tagged release. The project has been started since August 2025, and the coding agent VT Code itself I believe to be stable and ready to use. Thankfully I got helps from the community recently we pushed quite a lot of releases and enhancements and bug fixes.
VT Chat: https://github.com/vinhnx/vtchat > A modern, privacy-first AI chat application with security.
VT.ai: https://github.com/vinhnx/vt.ai > Multimodal AI chat app with dynamic conversation routing.
I missed favstar.fm so I tried making something similar on an open social network!
10 months is a long time. You can write one article per day and do it easily.
- a self-hosted PaaS with CI-driven deployments, managed by terraform (and possibly other IaC in the future) - think Coolify but with IaC or Komodo but easier to run full-stack apps (databases, queues and OTEL support built in)
- a benchmark for AI agents where they have to iteratively build an application - it shows how bad design decisions compound over time
Desperately trying to attract new monthly sponsors and people willing to buy me the occasional pizza with my terrible HTML skills. Is it working?
So I’m building an Android app designed to help me actually read what I save. You only get one saved article per day. Finish today’s read to unlock the next one. You can also save Twitter, Reddit, etc. posts separately and revisit them anytime.
Right now I’m mainly validating whether people would actually want something like this. If there’s enough interest, I’ll run a small closed beta for Android.
Join the waitlist for beta access: https://forms.gle/xgrXKoTkigYFcg6r7
I'm working on a procedural authoring system for party-based cRPGs. The idea is to treat RPG content as structured, validated data rather than loose text or one-off scripting. A procedurally generated location has encounters, hazards, NPCs, rewards, state changes, and a validation report. A generated quest has a hook, possible approaches, consequences, rewards, and failure states. Main campaign questlines, which can be modeled on the old "Adventure Path" system from Dungeon Magazine, are typically revealed in a stepwise way. The system automatically plays through and rejects content that is logically broken, unrewarding, or impossible to complete.
The broader goal is to build better authoring tools for RPGs: Tools that help small teams create large, reactive games very quickly, without giving up structure or design control.
So I'm starting a new company to push the boundaries of procedural generation and related technologies in games. No link to share yet. Maybe in a month or two. We do have playable games already, though.
It runs as a chrome extension and stores all the browser locally. A server is needed for some of the public connections. Early days but interesting:
I was frustrated with not being able to properly review code changes that agents make, looking for a middle ground between VSC and Claude Code. So I built one. Still in early development but its working (mostly) and I hope dogfooding will get me all the way to something great :)
At the end of the month, the app guides you through a reflection flow where you can learn about your habits and think about what you want to improve. 100% offline, fully private, no AI.
The app is paid, but I actually made a page for HN readers to get it for free, I'm having a lot of trouble with distribution: https://getkakeibo.com/hn/
Maybe add a link to playstore/appstore under the generated promo code
small things like this have outsized effect
Build complex Spotify playlists by wiring together components in a visual editor. Mix, filter, and sort tracks from any source (with Last.fm and Musicbrainz support). Combine multiple playlists, saved tracks, and recommendations. Filter by genre, audio features, popularity, and more. Schedule programs to keep playlists updated automatically.
I wrote about how I built the game’s dictionary and how hard the problem is: https://wordtrak.com/blog/2026-06-05-how-i-built-a-new-dicti...
The most surprising thing from the last month has been building a Discord app/activity has driven the most amount of engagement. Once it’s in your chat it tends to become a daily habit!
Give it a shot here: https://wordtrak.com/daily
Everyone is working on personal agents but their identity model is wrong. They act as you, risk your reputation, your data and more. Nym is a personal agent that has (and can make) all of its own accounts and only gets selective read only access to yours.
The goal is to make reliable agents that are able to operate safely in the world to help you do what you want, without exposing your accounts and personal identity to potential harms.
For instance nyms have their own e-mail addresses at nym-mail.com, you can CC them on chains and they can only respond to people on that chain with a lease of 5 days, or permanently for people you specifically add.
https://store.steampowered.com/app/4771830/Angel_Investor_Si...
You bid against other investors in a show similar to the Shark Tank and then you assign advisors to help the startup grow and finally cash out after IPO.
The ideas that startups have are wacky for the fun factor, but the economic simulation underneath is realistic.
There will be a free playable demo soon.
My primary project is Mermaid Studio, an IDE grade Mermaid diagram authoring experience for both humans and agents built for JetBrains IDEs. Semantic highlighting, refactoring, 120+ inspections covering everything from common rendering issues and foot-guns to obscure syntax quirks, contextual completions, interactive (drag-and-drop) diagram editing and styling. A suite of included MCP tools surfaces the plugin's intelligence and rendering capabilities, so your preferred agent (even using relatively small local models) can generate correct diagrams, even the newest or more obscure ones with ease.
https://mermaidstudio.dev/ | https://plugins.jetbrains.com/plugin/29870-mermaid-studio
Tinkering with several other IntelliJ plugin ideas that may or may not end up being released.
Still occasionally pushing up new features Tachi Code, my browser extension that brings a feature rich, highly customizable, Monaco Editor based code editor to the browser surfaces you always wished would be less of a pain to read. These days, I mostly use it for reading and printing Markdown files.
https://apps.apple.com/app/trace-on-device-transcripts/id676...
We map technical debt, agent readiness, code complexity, security scanning, bus factor so you can easily see how all the software in your company runs.
Got the top spot in What Are You Working On in Feb 2025 and then started a company around it at https://getcomper.ai . We got a pre-seed investment, then got some ex-Miro folk on board and are building the product at breakneck speed.
One of the coolest things is that we've got an incremental git blame engine built on top of GitOxide as our backend is fully built on Rust.
We are building local (Baltics) marketplace for teambuilding events.
Ever since I took on teamlead responsibilities in my daily job, it's been a recurring challenge to organize fun and engaging activities for a team. Providers are usually small businesses, so they all handle enquires, booking and payments to the best of their ability. That usually means exhausting back and forth to get and compare multiple offers.
Instead, we provide curated list of activity providers to choose from, and then handle communication, invoicing and disputes\returns, if needed.
We have got overwhelmingly good feedback from HRs, responsible for team event bookings, and now trying to grow vendors to launch publicly.
We've got: a language that looks like python but lowers to cpp: https://github.com/wegfawefgawefg/dudu; a site that tracks power outtages in canada: https://www.outagehub.ca/ ; a clone of spelunky in cpp with multiplayer and mods: https://github.com/wegfawefgawefg/splonks-cpp; a shadertoy clone but you code in asm: https://github.com/wegfawefgawefg/asm-shader-toy; a fixed point math lib: https://github.com/wegfawefgawefg/gfxp; a sexpr based json replacement with good perf: https://github.com/wegfawefgawefg/gsexp;
Outside of personal stuff, I'd like to shill my consultancy, G&K Software. (IDK if this is even allowed, but its what i'm working on so I think its fair.) We had: an ai asset generation site; a trading tool for a fund in dubai; an endorsed sales data tool;
Those are all out of business now due to global market craziness, tarrifs, and AI advancements... so I guess we've got nothing right now. (Please somebody give us business lol) gibson@gnk.software
It's a weekly email with all the recently published software engineering conference talks. I also pick a few ones that are featured and write short TLDRs. This month, I'll should hit 10,000 readers, fingers crossed!
ifh-hn•2h ago