Dev Systems - Building Systems That Scale

Show HN: I run a team of AI agents on my Kubernetes cluster

I built Axon, an open-source Kubernetes controller that turns AI coding agent runs into a declarative API. I can define my development workflow in YAML so that Axon handles the rest — isolation, credentials, git workspace, agent plugins, output capture.I use it to develop Axon itself. I run a team of specialized agents on my cluster, each with a different job:Workers — Watch for GitHub issues, pick them up, investigate, write code, open PRs, self-review, and iterate until CI passes. If an agent

Show HN: Sgai – Goal-driven multi-agent software dev (GOAL.md → working code)

Hey HN,We built Sgai to experiment with a different model of AI-assisted development.Instead of prompting step-by-step, you define an outcome in GOAL.md (what should be built, not how), and Sgai runs a coordinated set of AI agents to execute it.- It decomposes the goal into a DAG of roles (developer → reviewer → safety analyst, etc.) - It asks clarifying questions when needed - It writes code, runs tests, and iterates - Completion gates (e.g. make test) determine when it's actually doneEver

Show HN: Safari-CLI – Control Safari without an MCP

Hello HN!<p>I built this tool to help my agentic software development (vibe coding) workflow.<p>I wanted to debug Safari specific frontend bugs using copilot CLI, however MCP servers are disabled in my organisation. Therefore I built this CLI tool to give the LLM agent control over the browser.<p>Hope you'll find it useful!

Show HN: Ctrl and Click on text to jump to its translation key in VSCode

I built a small Chrome extension to make working with i18next translation keys easier during development.In larger projects, I often found myself searching through multiple JSON files to figure out which key was rendering a specific piece of text. Global search works, but it’s slow and sometimes noisy, especially when keys are reused or structured deeply.The idea was inspired by LocatorJS, which lets you click a component in the browser and jump to its source file. I liked that workflow and want

Show HN: TMDD – continuous threat modelling that makes your code more secure

My name is Mikolaj and I built tmdd tool, a CLI that keeps an up-to-date threat model of your app (in YAML format) in your repo and generates security-aware prompts for AI coding agents. Here's why:I am a security engineer. Oftentimes I observed situations where "technical" security levels were pretty high, but the business logic and authorization related issues made the apps pretty vulnerable. In my experience, no SAST or DAST is able to spot this kind of issues; some of them mig

Show HN: GalataJ – Runtime-aware Java profiling inside IntelliJ and VS Code

Hi HN,While working on large Java projects, I often found myself digging through heap dumps and flame graphs just to answer a simple question:Which methods are actually slow, and how much do they cost?Traditional profilers are powerful, but interpreting the output takes time. AI tools can read source code — but they don’t see what happens at runtime.So I built GalataJ.GalataJ is a lightweight Java profiler that works directly inside IntelliJ (including Community Edition) and VS Code. The workflo

Show HN: VaultSandbox – Real SMTP testing that works on localhost (Apache 2.0)

Hi HN,I posted VaultSandbox two months ago and the main feedback was AGPL concerns. I've listened: it's now Apache 2.0. I've also overhauled the local experience to make it "just work" on localhost with zero config.VaultSandbox is a self-hosted SMTP receiver and programmable debugger. It bridges the gap between "it sent" and "it was delivered" by simulating production-grade email lifecycles. It is architected for parallel testing: each test runs in it

Show HN: Made First Android game using Codespaces and AI, now in AdMob purgatory

Two months ago, I had zero mobile development or publishing experience. I wanted to see if I could build and ship a complete game using only a browser-based workflow and AI assistance.The process:Built entirely in GitHub Codespaces using an AI Copilot.Wrote the core game for the web, then wrapped it into an Android APK/AAB.Pushed through 100+ versions to fix edge cases.Passed Google Play review with no policy violations.The game is a tough 2D platformer called Bionic Biome. To make the &quo

Show HN: Emdash – Open-source agentic development environment

Hey HN! We’re Arne and Raban, the founders of Emdash (https://github.com/generalaction/emdash).Emdash is an open-source and provider-agnostic desktop app that lets you run multiple coding agents in parallel, each isolated in its own git worktree, either locally or over SSH on a remote machine. We call it an Agentic Development Environment (ADE).You can see a 1 minute demo here: https://youtu.be/X31nK-zlzKoWe are building Emdash for ourselves. While working on a

Show HN: Agent workspace switcher for Git worktrees

Hey HN, I've been exploring ways to make my engineering teams more productive in multi-agent AI coding workflows.The most popular development tools (Cursor, Claude Code, etc.) leverage git worktrees to enable multiple agents to work in parallel, guaranteeing each runs in a semi-isolated workspace. The workflow is powerful, but switching between worktrees to test what each agent built is painful: take down the stack for agent X, cd to the right worktree, re-source .env files, bring up the st

Digital Transformation at Santander: How Platform Engineering is Revolutionizing Cloud Infrastructure

This post is cowritten by Julio Bando from Santander. Santander faced a significant technical challenge in managing an infrastructure that processes billions of daily transactions across more than 200 critical systems. The expansion into diverse financial services, including investment banking, wealth management, insurance, and payment solutions, had created unprecedented technological complexity, requiring a robust, agile, and scalable infrastructure solution. This raised two main issues. Santa

Show HN: Chaos Monkey but for Audio Video Testing (WebRTC and UDP)

It takes an input video and converts it into H.264/Opus RTP streams that you can blast at your video call systems (WebRTC, SFUs, etc.). It also injects network chaos like packet loss, jitter, and bitrate throttling to see how things breakIt scales from 1 to n participants, depending on the compute and memory of the host system Best part? It’s packaged with Nix, so it builds the same everywhere (Linux, macOS, ARM, x86). No dependency hellIt supports both UDP (with a relay chain for Kubernete

Show HN: PgDog – Scale Postgres without changing the app

Hey HN! Lev and Justin here, authors of PgDog (https://pgdog.dev/), a connection pooler, load balancer and database sharder for PostgreSQL. If you build apps with a lot of traffic, you know the first thing to break is the database. We are solving this with a network proxy that works without requiring application code changes or database migrations.Our post from last year: https://news.ycombinator.com/item?id=44099187The most important update: we are in production. S

Ask HN: Why is network documentation still a scavenger hunt for the truth?

After nearly two decades around DC networking (design, maintenance, migrations), the consistent problem I see isn’t “how to get configs” it’s drift and it feels like it’s getting worse as architectures become more distributed. Whatever you call the “source of truth” (DCIM/IPAM/CMDB, docs/spreadsheets, quick sketches in gimp/mspaint, diagrams, GIS/KMZ, vendor dashboards) almost always diverges from the running state. The first thing to decay is usually the relationships:

Show HN: I built an AI senior architect – vibe coding meets system design

I've been designing distributed systems for years, for work, interviews, side projects. I wanted one place to do all of it: refresh my architecture knowledge, keep up with how companies like Netflix and Stripe actually build things, and try out design ideas without starting from a blank page every time.So I built SysDesAi. It started as a tool for myself. Part AI architect, part learning platform, part engineering news source. You describe what you want to design and an AI walks through it

Show HN: Opty – A Zig-based HDC that reduces token use by up to 90%

I kept seeing people all over social media talking about how they were making custom local-LLM systems to help reduce the token load injected into their context window.On the side my recent project has me looking at Hyperdimensional Distributed Memory. I couldn't help but wonder if I could make an MCP server to improve token usage. Sure enough, using a combination of HDC + TOOL format, I was able to get opty's own self audit down by 93% in token usage.Still experimenting with large cod

A thought on quantum error correction: accuracy without replay feels fragile

I’ve been thinking about quantum error correction work lately, and one thing keeps bothering me.A lot of effort goes into reporting decoder accuracy improvements, but much less into whether those results are replayable over time or safe to compare after assumptions change.In practice, small shifts in noise behavior, detector mapping, or measurement stability can quietly invalidate earlier conclusions. Often everything still “looks reasonable,” so regressions go unnoticed until much later.It feel

Show HN: AI-Native "Medicare for All" Prototype

Medicare.dev is a fully-functional prototype of how I think an AI-native, single-payer healthcare system could work. It uses "programs" and not "plans" because I do not think plans make sense given that nobody plans on getting sick. It also uses one big market-network instead of provider networks, you can also add your own providers if you want to.How It Works:1. Define the Problem — A patient or provider describes a health need in natural language. Just press the mic and ex

Show HN: Raypher–Sandboxing local AI agents(OpenClaw)on your own local computer

Hey HN,Most of us want to run autonomous agents like OpenClaw locally, right on our daily-driver machines. We want them to actually interact with our files, our IDEs, and our real local workflows.But doing that right now is a security nightmare. A hallucinating (or hijacked) agent with raw system access is basically local remote code execution waiting to wipe a directory or leak an SSH key.The current workarounds are terrible:The Hardware Air-Gap: Buying a dedicated Mac Mini or a secondary cheap

Ask HN: Does anyone know how to register as a developer at Microsoft?

I'm trying to distribute an Office Add-in. The Partner Center developer/company profile confirmation system is completely broken (trying to avoid flame bait) and Microsoft Support won't pick up the phone or chat. I've been stuck for over a month now, does anyone know the way through this process?