🛠️ All DevTools

[Other] Show HN: Browser4 – an open-source browser engine for agents and concurrency Hi HN,<p>I’d like to share an open-source project we’ve been working on for a while: <i>Browser4</i>.<p>The motivation came from a recurring frustration: most browser automation tools (Playwright, Selenium, Puppeteer) are excellent for <i>human-written scripts</i>, but start to show friction when used as a <i>core execution layer for AI agents</i> or at very high concurrency.<p>So instead of building “another wrapper around Playwright”, we experimented with a different direction: <i>designing a browser engine where AI agents are first-class citizens.</i><p>### What Browser4 is<p>Browser4 is a browser automation engine built on <i>native Chrome DevTools Protocol (CDP)</i>, with a focus on:<p>* <i>Coroutine-safe concurrency</i> (designed to run many browser sessions in parallel)<p>* <i>Agent-oriented APIs</i> (navigation, interaction, extraction as composable actions)<p>* <i>Hybrid extraction</i>: ML agent driven extraction + LLM extraction + structured selectors + an SQL-like DOM query language (X-SQL)<p>* <i>Low-level control</i> without Playwright-style abstraction overhead<p>It’s written in <i>Kotlin&#x2F;JVM</i>, mainly because we needed predictable concurrency behavior and long-running stability under load.<p>The project is fully open-source (Apache 2.0).<p>### What it’s <i>not</i><p>* It’s not a drop-in Playwright replacement.<p>* It’s not a no-code RPA tool.<p>* It’s not “LLM magic” — LLMs sit <i>outside</i> the browser engine.<p>Browser4 intentionally stays close to the browser execution layer and leaves planning&#x2F;reasoning to external agent loops.<p>### Current use cases we’re testing<p>* Large-scale web data extraction<p>* Agentic workflows (search → navigate → extract → summarize)<p>* Price &#x2F; content monitoring with frequent revisits<p>* High-concurrency crawling where browser startup and context switching are bottlenecks<p>On a single machine, we can sustain <i>very high daily page visits</i>, though we’re still validating benchmarks across different workloads.<p>### Open questions (where I’d love feedback)<p>* For agentic systems, does it make sense to bypass Playwright entirely and work closer to CDP?<p>* Where do you see the biggest pain points when combining LLMs with browser automation today?<p>* Is JVM a reasonable choice here, or is Python still the better tradeoff despite concurrency limits<p>* What abstractions would <i>you</i> want in a browser engine built for AI agents?<p>### Links<p>* GitHub: <a href="https:&#x2F;&#x2F;github.com&#x2F;platonai&#x2F;browser4" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;platonai&#x2F;browser4</a><p>* Website (light overview): <a href="https:&#x2F;&#x2F;browser4.io" rel="nofollow">https:&#x2F;&#x2F;browser4.io</a><p>Happy to answer technical questions or hear criticism — especially from people running browser automation or agent systems in production.<p>Thanks for reading.

[Database] Show HN: EdgeVec – Sub-millisecond vector search in the browser (Rust/WASM) Hi HN,<p>I built EdgeVec, a vector database that runs entirely in the browser. It implements HNSW (Hierarchical Navigable Small World) graphs for approximate nearest neighbor search.<p>Performance: - Sub-millisecond search at 100k vectors (768 dimensions, k=10) - 148 KB gzipped bundle - 3.6x memory reduction with scalar quantization<p>Use cases: browser extensions with semantic search, local-first apps, privacy-preserving RAG.<p>Technical: Written in Rust, compiled to WASM. Uses AVX2 SIMD on native, simd128 on WASM. IndexedDB for browser persistence.<p>npm: <a href="https:&#x2F;&#x2F;www.npmjs.com&#x2F;package&#x2F;edgevec" rel="nofollow">https:&#x2F;&#x2F;www.npmjs.com&#x2F;package&#x2F;edgevec</a> GitHub: <a href="https:&#x2F;&#x2F;github.com&#x2F;matte1782&#x2F;edgevec" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;matte1782&#x2F;edgevec</a><p>This is an alpha release. Main limitations: build time not optimized, no delete operations yet.<p>Would love feedback from the community!

[Other] Show HN: I built a GitHub application that generates documentation automatically Hi HN,<p>A lot of the dev teams I have worked with had a lot of issues with their documentation. In fact, some of my easiest clients to get were from clients that had &quot;black box&quot; solutions that devs no longer at the company had created. Personally, writing documentation is like grinding nails on a chalkboard.<p>I have been having a lot of fun with building solutions that can run in a distributed way, not something a dev needs to run themselves. And after a significant amount of testing and building out several different solutions, I finally have a solution that is easy to set up and runs in the background continuously to automate the documentation process.<p>I&#x27;m looking for feedback on a few things: - Ease of onboarding, it should be a simple click -&gt; select repos you want to add. - Quality of documentation, our current free accounts have a standard model compared to premium but the concepts are the same. - Dynamic environments: I tried to make this compatible with any random repo thrown at it.<p>Let me know your thoughts

[API/SDK] Show HN: Verani – Socket.io-like realtime SDK for Cloudflare Hi HN,<p>I built Verani, an experimental realtime SDK for Cloudflare Actors (Durable Objects).<p>The goal is to bring a familiar, event-based DX (rooms, emit, channels, lifecycle hooks) to Cloudflare Durable Objects, while handling hibernation correctly.<p>It focuses on - Socket.io-like semantics on top of Durable Objects - Proper Actor hibernation handling (WebSockets survive, state is restored) - Typed API with explicit tradeoffs.<p>It’s early and experimental, but functional. I built it mainly to explore DX and state management patterns on Cloudflare Actors, and I’d love feedback from people who’ve built realtime systems or used Durable Objects.

[Other] I couldn't find a logging library that worked for my library, so I made one

[Database] Show HN: Marmot v2.20 – A distributed SQLite server with MySQL wire compatbility Marmot hits major milestone I&#x27;ve been solo pushing (with help of AI), to really mature out the system. And yesterday I have hit the first milestone where MySQL API is stable enough to make WordPress install and work flawlessly. System is now in a place where you can spin off a cluster, and then spin off many replicas you want to scale out. I wanted to build something that is actually hard with AI, and it&#x27;s been long and tough journey trying various AI tools, and huge amount of learning. I will follow up with a blog post, but since I have all the scripts to help you guys spin up cluster and examples I wanted to put it out there so that you guys can poke holes and help me improve it.<p>Source is available at: <a href="https:&#x2F;&#x2F;github.com&#x2F;maxpert&#x2F;marmot&#x2F;" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;maxpert&#x2F;marmot&#x2F;</a><p>Most fun part is a Wordpress running in cluster: <a href="https:&#x2F;&#x2F;github.com&#x2F;maxpert&#x2F;marmot&#x2F;tree&#x2F;master&#x2F;examples&#x2F;wordpress-cluster" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;maxpert&#x2F;marmot&#x2F;tree&#x2F;master&#x2F;examples&#x2F;wordp...</a><p>Each one of these Wordpress instances is talking to it&#x27;s own &quot;MySQL&quot; but under the hood replication is being done and stored over SQLite DB.

[Database] Turso is an in-process SQL database, compatible with SQLite.

[Other] Building small Docker images faster

[Build/Deploy] Build cloud-native AI Agents Deploy and manage fleets of AI agents and sub-agents as reliably as modern microservices. Built on an open-source golang SDK, run your agents anywhere (localhost, your cloud or ours). You own all of the code, quick deploy in under 2 minutes, Supporting infra and tools (data stores, vector db, memory, etc) included, generous freemium tier to help you get started with something real.

[Other] Playful app idea generator for web/mobile developers What To Build Next is a playful, AI-powered idea wizard that gives developers personalized app ideas based on their goals, skills, interests, and tools. Instead of generic idea lists, it asks thoughtful questions and generates ideas that actually fit you. Each idea includes your chosen intention, industry and tech stack so you can stop scrolling for inspiration and finally start developing.

[Database] Stoolap: High-performance embedded SQL database in pure Rust

[CLI Tool] Show HN: Gotui – a modern Go terminal dashboard library 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.

[Other] Show HN: I built a WebMIDI sequencer to control my hardware synths Hey HN,<p>I’m an ex-Google engineer trying to get back into music production.<p>I needed a way to sequence my hardware synths using AI contexts without constantly switching windows, so I built this.<p>It runs entirely in the browser using WebMIDI. No login required. It connects to your local MIDI devices (if you&#x27;re on Chrome&#x2F;Edge) and lets you generate patterns.<p>Tech stack: [React &#x2F; WebMIDI API &#x2F; etc].<p>Link: www.simplychris.ai&#x2F;droplets<p>Code is a bit messy, but it works. Feedback welcome.

[Other] Days since last GitHub incident

[CLI Tool] Show HN: Chess-TUI Play Lichess games in your terminal (Rust) Hi HN, I’ve released a new version of chess-tui, a Rust terminal UI for playing live games against Lichess opponents.<p>The update includes a cleaner TUI, better performance, improved keybindings, and smoother real-time integration with the Lichess API. It’s designed as a simple, fast way to play chess without leaving the terminal.<p>Repo:<a href="https:&#x2F;&#x2F;github.com&#x2F;thomas-mauran&#x2F;chess-tui" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;thomas-mauran&#x2F;chess-tui</a><p>Feedback and bug reports are welcome. Happy to answer questions!

[DevOps] Ship AI Agents to Google Cloud in minutes, not months. Production-ready templates with built-in CI/CD, evaluation, and observability.

[DevOps] Show HN: Titan – JavaScript-first framework that compiles into a Rust server Hi HN,<p>I built Titan, a backend framework where you write routes and logic in JavaScript, and the CLI compiles everything into a single Rust + Axum binary using the Boa JS engine. No Node.js is required in production.<p>The idea is to keep JS developer experience while getting Rust performance and a self-contained deployable server.<p>Current features:<p>JS route DSL<p>Action system mapped to Rust<p>esbuild bundling<p>Generated Rust server with Axum<p>Hot-reload dev server<p>Single-binary output<p>Repo: <a href="https:&#x2F;&#x2F;github.com&#x2F;ezet-galaxy&#x2F;-ezetgalaxy-titan" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;ezet-galaxy&#x2F;-ezetgalaxy-titan</a><p>Would love feedback on the architecture, DX, and whether this hybrid JS→Rust approach is useful.<p>Thanks for reading!

[API/SDK] A modern data client for React & tRPC fate is a modern data client for React and tRPC inspired by Relay and GraphQL. It combines view composition, normalized caching, data masking, Async React features, and tRPC's type safety. fate is designed to make data fetching and state management in React applications more composable, declarative, and predictable. The framework has a minimal API, no DSL, and no magic—it's just JavaScript.

[CLI Tool] State-of-the-art Devstral 2: 123B-parameter open-source coding model achieving 72.2% on SWE-bench Verified. Mistral Vibe CLI: Native coding assistant solving software engineering tasks.

[Other] All-in-one local data converter. Many tools, 100% privacy. Free, secure, and privacy-focused developer tools. Convert CSV, JSON, SQL, YAML, and contacts locally in your browser. No server-side processing.