🛠️ All DevTools

[Other] Show HN: Build the habit of writing meaningful commit messages Too often I find myself being lazy with commit messages. But I don&#x27;t want AI to write them for me... only i truly know why i wrote the code i did.<p>So why don&#x27;t i get AI to help me get that into words from my head?<p>That&#x27;s what i built: smartcommit asks you questions about your changes, then helps you articulate what you already know into a proper commit message. Captures the what, how, and why.<p>Built this after repeatedly being confused 6 months in a project as to why i made the change i had made...<p>Would love feedback!

[Other] Show HN: I turned algae into a bio-altimeter and put it on a weather balloon Hi HN - My name is Andrew, and I&#x27;m a high school student.<p>This is a write-up on StratoSpore, a payload I designed and launched to the stratosphere. The goal was to test if we could estimate physical altitude based on algae fluorescence (using a lightweight ML model trained on the sensor data).<p>The blog post covers the full engineering mess&#x2F;process, including:<p>- The Hardware: Designing PCBs for the AS7263 spectral sensor and Pi Zero 2 W.<p>-The biological altimeter: How I tried to correlate biological stress (fluorescence) with altitude.<p>- The Communications: A custom lossy compression algorithm I wrote to smash 1080p images down to 18x10 pixels so I could transmit them over LoRA (915 Mhz) in semi-real-time.<p>The payload is currently lost in a forest, but the telemetry data survived. The code and hardware designs are open source on GitHub: <a href="https:&#x2F;&#x2F;github.com&#x2F;radeeyate&#x2F;stratospore" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;radeeyate&#x2F;stratospore</a><p>I&#x27;m happy to answer technical questions about the payload, software, or anything else you are curious about! Critique also appreciated!

[CLI Tool] Detect process injection and memory tampering in Rust Built Ghost - scans processes for signs of malware injection. Catches shellcode, API hooks, process hollowing, thread hijacking, that stuff. Works on Windows, Linux, macOS. Pretty fast, scans 200 processes in about 5 seconds. Has both command line and terminal UI. Fair warning - you'll get false positives from browsers and game anti-cheat because they do weird memory stuff. So don't freak out if it flags Chrome. Open source, MIT license. Drop a star if you find it useful.

[API/SDK] Search and Alert on Livestream Content Data Core.stream delivers search and alerting on livestream content data. We are API first, enabling builders to create cool content. Our algorithm picks out consistent streamers with audiences and transcribes them using data from their live feed. We made a demo app of what you can build on us at whoistalkingabouttacos.com, fully using the core.stream api. Useful for developers, brand managers, searching for content trends, but can have wide-reaching applications.

[Other] Your swiss army knife for productivity 115+ utility tools in one platform — AI content creators, developer utilities, image processors, and data converters. From AI blog generation to code formatting. Everything you need, instantly accessible.

[Other] Building the largest known Kubernetes cluster

[CLI Tool] Show HN: Transcribe Your Voice in Terminal Locally Use hns, a speech-to-text CLI tool to transcribe your voice from your microphone directly to clipboard. Integrate hns with Claude Code, Ollama, LLM, and more CLI tools for powerful workflows.<p>hns transcribes your voice 100% locally using faster-whisper. The whisper model is downloaded automatically on first run and after that, hns can be used completely offline. After transcription, the text is displayed in the terminal (written to stdout) as well as automatically copied to your clipboard, ready to be pasted anywhere with Ctrl+V or Cmd+V.<p>GitHub: <a href="https:&#x2F;&#x2F;github.com&#x2F;primaprashant&#x2F;hns&#x2F;" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;primaprashant&#x2F;hns&#x2F;</a>

[Other] Show HN: OCR Arena – A playground for OCR models I built OCR Arena as a free playground for the community to compare leading foundation VLMs and open-source OCR models side-by-side.<p>Upload any doc, measure accuracy, and (optionally) vote for the models on a public leaderboard.<p>It currently has Gemini 3, dots.ocr, DeepSeek, GPT5, olmOCR 2, Qwen, and a few others. If there&#x27;s any others you&#x27;d like included, let me know!

[Other] Show HN: Wealthfolio 2.0- Open source investment tracker. Now Mobile and Docker Hi HN, creator of Wealthfolio here.<p>A year ago, I posted the first version. Since then, the app has matured significantly with two major updates:<p>1. Multi-platform Support: Now available on Mobile (iOS), Desktop (macOS, Windows, Linux), and as a Self-hosted Docker image. (Android coming soon).<p>2. Addons System: We added explicit support for extensions so you can hack around, vibe code your own integrations, and customize the app to fit your needs.<p>The core philosophy remains the same: Always private, transparent, and open source.

[Other] Building a Minimal Viable Armv7 Emulator from Scratch

[Other] Autocomp: An ADRS Framework for Optimizing Tensor Accelerator Code Paper: <a href="https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;2505.18574" rel="nofollow">https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;2505.18574</a>

[API/SDK] Show HN: Docuglean – Extract Structured Data from PDFs/Images Using AI Hi HN! I built Docuglean, an open-source SDK for intelligent document processing that works with OpenAI, Mistral, Google Gemini, and Hugging Face models.<p>The idea came from repeatedly writing boilerplate code to extract structured data from invoices, receipts, and other documents. Instead of wrestling with different API formats, I wanted a unified interface that:<p>- Extracts structured data using Zod&#x2F;Pydantic schemas - Classifies and splits multi-section documents (e.g., medical records) - Processes documents in batches with automatic error handling - Works locally without APIs (for PDFs, DOCX, XLSX, etc.)<p>Key features: - Available for both TypeScript and Python - Batch processing with concurrent requests - Document classification (splits 100+ page docs by category) - Local parsers (no API needed for basic extraction) - Apache 2.0 licensed<p>Currently supports OpenAI, Mistral, Gemini, and Hugging Face. Planning to add Together AI, Anthropic, and more.<p>Would love feedback on the API design and what features would be most useful

[DevOps] Run Docker containers natively in Proxmox 9.1 (OCI images)

[Other] Show HN: GitPulse – AI-powered tool to discover open source projects I built GitPulse to solve a problem I had: finding beginner-friendly repos.<p>Features: • 200+ curated “good first issues” • AI-powered difficulty predictor • Smart repo matching • Contributor analytics • Repo health score<p>Live: <a href="https:&#x2F;&#x2F;git-pulsee.vercel.app" rel="nofollow">https:&#x2F;&#x2F;git-pulsee.vercel.app</a>

[Testing] Show HN: Supabase-Test – Fast Isolated Postgres DBs for Testing Supabase RLS Hi HN — we&#x27;ve built a testing framework for Supabase that spins up fast, isolated Postgres databases for each test case. It’s designed to make RLS policies easy to validate with real database state, without global test fixtures or mock auth.<p>Features: - Instant isolated Postgres DBs per test - Automatic rollback after each test - RLS-native testing with `.setContext()` for auth simulation - Flexible seeding (SQL, CSV, JSON, JS) - Works with Jest, Mocha, and any async test runner - CI-friendly (runs cleanly in GitHub Actions)<p>We also published example projects and a free set of tutorials: <a href="https:&#x2F;&#x2F;launchql.com&#x2F;learn&#x2F;supabase" rel="nofollow">https:&#x2F;&#x2F;launchql.com&#x2F;learn&#x2F;supabase</a><p>Package: <a href="https:&#x2F;&#x2F;www.npmjs.com&#x2F;package&#x2F;supabase-test" rel="nofollow">https:&#x2F;&#x2F;www.npmjs.com&#x2F;package&#x2F;supabase-test</a><p>Source + full test suite: <a href="https:&#x2F;&#x2F;github.com&#x2F;launchql&#x2F;supabase-test-suite" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;launchql&#x2F;supabase-test-suite</a><p>Happy to answer questions and get feedback, cheers :)

[Other] Show HN: YAAT – Privacy-first analytics for EU companies (need for beta users) I built YAAT for EU companies that can&#x27;t of send data to US servers or being locked into pre-built dashboards that can&#x27;t answer custom questions.<p>What makes it different:<p>Direct SQL access to your data. Not just pre-built reports – write actual queries against your raw events. Want to know which UTM campaigns convert best for mobile users in Germany? Write the query, get the answer, save it as a dashboard panel.<p>Full analytics stack:<p>Web analytics: pageviews, sessions, traffic sources, UTM tracking, device&#x2F;browser&#x2F;geo data Error tracking: JavaScript exceptions, unhandled promises, stack traces, filtering by browser&#x2F;version Performance monitoring: Core Web Vitals (LCP, FID, CLS, INP), page load times, TTFB<p>Custom dashboards: Drag-drop panels anywhere. Time series, bar charts, pie charts, maps, tables. Monaco editor with SQL autocomplete. Export data as Parquet files – full ownership.<p>Privacy-first: EU-hosted infrastructure (no data transfers), GDPR-compliant, no cookies needed, lightweight script (&lt;2KB).<p>Domain verification via DNS ensures only your sites can send data.<p>Current state: Beta with 7 verified domains tracking production traffic. Looking for 10 EU companies to test for 3 months free.<p>Want feedback on SQL interface and what analytics patterns matter most for your business.<p>Try it: <a href="https:&#x2F;&#x2F;yaat.io&#x2F;beta" rel="nofollow">https:&#x2F;&#x2F;yaat.io&#x2F;beta</a><p>Built in Valencia, Spain. All data stays in EU.

[Other] Show HN: Tangent – Security log pipeline powered by WASM Hi HN! We’re Ethan and Danny, the authors of Tangent (<a href="https:&#x2F;&#x2F;github.com&#x2F;telophasehq&#x2F;tangent" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;telophasehq&#x2F;tangent</a>), a Rust-based log pipeline where all normalization, enrichment, and detection logic runs as WASM plugins.<p>We kept seeing the same problems in the OCSF (<a href="https:&#x2F;&#x2F;ocsf.io" rel="nofollow">https:&#x2F;&#x2F;ocsf.io</a>) community: 1) Schemas change constantly. Large companies have whole teams dedicated to keeping vendor→OCSF mappings up to date. 2) There’s no shared library of mappings, so everyone recreates the same work. 3) Writing mappers is tedious, repetitive work. 4) Most pipelines use proprietary DSLs that are hard to share and hard for tools&#x2F;LLMs to generate.<p>Tangent takes a different approach: no DSLs – mappings and enrichments are just normal code compiled to WASM, shareable plugins – we maintain a community library (<a href="https:&#x2F;&#x2F;github.com&#x2F;telophasehq&#x2F;tangent-plugins" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;telophasehq&#x2F;tangent-plugins</a>), interoperability – we can run other engines’ DSLs (e.g., Bloblang) inside WASM for easy migration, full flexibility – plugins can validate schemas, call external APIs (<a href="https:&#x2F;&#x2F;github.com&#x2F;telophasehq&#x2F;tangent&#x2F;blob&#x2F;main&#x2F;examples&#x2F;enrichment&#x2F;main.go#L58" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;telophasehq&#x2F;tangent&#x2F;blob&#x2F;main&#x2F;examples&#x2F;en...</a>), or perform complex transforms (<a href="https:&#x2F;&#x2F;github.com&#x2F;telophasehq&#x2F;tangent-plugins&#x2F;blob&#x2F;main&#x2F;zeek-ocsf&#x2F;conn&#x2F;main.go#L312" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;telophasehq&#x2F;tangent-plugins&#x2F;blob&#x2F;main&#x2F;zee...</a>).<p>Here&#x27;s an example Python transformation plugin to drop all fields from a log except `message`:<p><pre><code> import json from typing import List from wit_world.imports import log # `log.Logview` is Tangent&#x27;s zero-copy JSON accessor type. def process_logs(self, logs: List[log.Logview]) -&gt; bytes: out = bytearray() for lv in logs: msg = lv.get(&quot;msg&quot;) value = msg.value if msg is not None else &quot;&quot; out.extend(json.dumps({&quot;message&quot;: value}).encode() + b&quot;\n&quot;) return bytes(out) </code></pre> We have plenty more examples in the repo.<p>Because plugins are just Go&#x2F;Python&#x2F;Rust, LLMs can create new mappers with ease. For example, I asked:<p><pre><code> Generate a mapper from AWS Security Hub Finding to OCSF </code></pre> and only had to make a few minor tweaks. (<a href="https:&#x2F;&#x2F;github.com&#x2F;telophasehq&#x2F;tangent-plugins&#x2F;blob&#x2F;main&#x2F;aws-ocsf&#x2F;security_hub&#x2F;main.go" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;telophasehq&#x2F;tangent-plugins&#x2F;blob&#x2F;main&#x2F;aws...</a>)<p>Performance-wise, a 16-core Amazon Linux box processes ~480 MB&#x2F;s end-to-end (TCP → Rust-WASM transform → sink) on ~100-byte JSON logs. The CLI includes tooling to scaffold, test, and benchmark plugins locally. Here&#x27;s a deep dive into how we are able to get this performance: <a href="https:&#x2F;&#x2F;docs.telophasehq.com&#x2F;runtime" rel="nofollow">https:&#x2F;&#x2F;docs.telophasehq.com&#x2F;runtime</a>.<p>We’d love to get your feedback! What do you think?

[Other] Show HN: Awesome J2ME An awesome list about Java platform Micro edition(J2ME). Documentation, academic papers, tutorials, communities, IDEs, SDKs, emulators, apps, video games. J2ME is a Java specification designed for old keypad phones and PDAs. MIDP, which is built upon CLDC, is used to create Midlets, which have `.jad` or `.jar` extension, and run on platforms like old keypad phones, Symbian and PDAs. MIDP is supported till Java ME SDK 3.4.

[Other] Show HN: Lamina – A compiler backend that is not LLVM or Cranelift Recently, I&#x27;ve been working on Lamina, a compiler infrastructure that generates native assembly for multiple architectures without relying on LLVM or Cranelift. It&#x27;s designed for building compilers for new languages, educational projects, and any projects that can utilize a custom syntax of code generation.<p>Instead of depending on external backends, Lamina provides a complete pipeline from a single SSA based IR directly to the supported target&#x27;s assembly generation. The IR is readable, also provides a IRBuilder API that is easy to use via programmatic construction.<p>For better management of the code generation process, in the future, it will use a new pipeline IR -&gt; MIR -&gt; native assembly with the optimization passes.<p>Key features: - Direct code generation: IR -&gt; assembly&#x2F;machine code without LLVM&#x2F;Cranelift - SSA based IR: single assignment form optimized for analysis and optimization passes - MIR based codegen(experimental): new intermediate representation with register allocation and advanced optimizations - IRBuilder API: fluent interface for building modules, functions, blocks, and control flow - Readable IR: easy to debug and lower than high level languages - Zero external backend dependencies: simplified builds and transparent pipeline while being faster to build<p>Optimization passes (experimental MIR flow only): - Control flow: CFG simplification, jump threading, branch optimization - Loop optimizations: loop fusion, loop invariant code motion, loop unrolling - Code motion: copy propagation, common subexpression elimination, constant folding - Function optimizations: inlining, tail call optimization - Arithmetic: strength reduction, peephole optimizations<p>Performance: On a 256×256 matrix multiplication benchmark (300 runs), Lamina&#x27;s experimental MIR-based codegen (which includes all optimization passes) generates code comparable to C&#x2F;C++&#x2F;Rust (within 1.8x) and faster than Java, Go, JavaScript, and Python. The experimental MIR based flow&#x27;s result is much faster than the IR-&gt; Assembly based codegen.<p>Written in Rust (2024 edition), Current Version 0.0.7. Optional nightly features available for SIMD, atomic placeholders, and experimental targets.

[Other] Show HN: CTON: JSON-compatible, token-efficient text format for LLM prompts