IDF killed Gaza aid workers at point blank range in 2025 massacre: Report

https://substackcdn.com/image/fetch/$s_!d2dB!,w_5760,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F174f0566-d8ca-4cea-a4c7-360a4f016881_4000x3000.jpeg
Israeli soldiers fired nearly 1,000 bullets at 15 Palestinian aid workers in southern Gaza on March 23, 2025, killing them in a massacre. The investigation by Earshot and Forensic Architecture found that the soldiers executed some aid workers at point blank range and that the Israeli military's internal inquiry did not recommend any criminal action.

I'm helping my dog vibe code games

https://www.calebleak.com/blog/dog-game/screenshot_20260219_154701.webp
The user trained their 9-pound cavapoo Momo to create games using Claude Code by routing her keyboard input through a Raspberry Pi and providing automated feedback tools. With a well-crafted prompt and strong guardrails, Momo's keyboard mashing produced playable games, demonstrating that the bottleneck in AI-assisted development is the quality of feedback loops, not the quality of ideas.

OpenAI, the US government and Persona built an identity surveillance machine

https://vmfunc.re/persona/shodan02.png
The authors of this document conducted a passive reconnaissance investigation into OpenAI's identity verification platform, Persona, and found that it serves facial recognition, watchlists, and SAR filings on a government endpoint with the door wide open. The platform's source code, which was publicly served without authentication, reveals a KYC/AML compliance platform that handles sensitive ...

Mac mini will be made at a new facility in Houston

https://www.apple.com/newsroom/images/2026/02/apple-accelerates-us-manufacturing-with-mac-mini-production/article/Apple-US-manufacturing-investment-Houston-data-center-assembly-line-01_big.jpg.large.jpg
Apple is expanding its Houston factory operations, creating thousands of jobs and producing Mac mini in the US for the first time. The company will also expand AI server manufacturing and open an Advanced Manufacturing Center for hands-on training.

Pi – A minimal terminal coding harness

https://pi.dev/tree-view.png
Pi is a minimal terminal coding harness with powerful defaults and extensibility. It adapts to workflows with TypeScript extensions, skills, and themes, and can be extended or installed with third-party packages.

I pitched a roller coaster to Disneyland at age 10 in 1978

https://wordglyph.xyz/Quadrupuler-Blueprints.jpg
The author, a 10-year-old boy, invented a roller coaster model called the Quadrupuler with four loops and sent it to Disneyland, receiving a positive response from Disney Imagineering. This early validation gave him a resilient spirit that has guided him through various inventions and a career in acting.

How we rebuilt Next.js with AI in one week

https://cf-assets.www.cloudflare.com/zkvhlag99gkb/64oL10LCSz30EiEHWVdJPG/aeaed48a681ec3d5bc9d8cb6e5e30a96/BLOG-3194_1.png
Cloudflare engineers rebuilt Next.js from scratch using AI and Vite, creating vinext, a drop-in replacement that deploys to Cloudflare Workers with a single command. Vinext builds production apps up to 4x faster and produces client bundles up to 57% smaller.

Firefox 148 Launches with AI Kill Switch Feature and More Enhancements

https://serverhost.com/blog/wp-content/uploads/2026/02/4fa59d91-110e-42c3-b1fa-e372279736ab.webp
Firefox 148 introduces an AI kill switch and enhanced user controls, allowing users to disable AI features and opt out of remote updates. The update also includes various performance and accessibility improvements, such as XSS protection and screen reader compatibility.

Open Letter to Google on Mandatory Developer Registration for App Distribution

https://keepandroidopen.org/img/logos/adguard.com.png
We, civil society and tech organizations, strongly oppose Google's new policy requiring Android app developers to register with Google to distribute apps outside the Google Play Store. This policy threatens innovation, competition, and user freedom by giving Google too much control over the Android platform.

Blood test boosts Alzheimer's diagnosis accuracy to 94.5%, clinical study shows

This page will redirect in a moment...

Discord cuts ties with identity verification software, Persona

https://fortune.com/img-assets/wp-content/uploads/2026/02/GettyImages-1239812755-e1771937537913.jpg?format=webp&w=1440&q=100
Discord's identity verification software, Persona, had frontend code accessible on the open internet and government servers, exposing sensitive user information. Discord has since cut ties with Persona, citing a short-lived partnership that lasted less than a month.

Nearby Glasses

https://opengraph.githubassets.com/6eb0426428f2b61bb8f668079806cf1ef864def5ff2f92cbf9b715a1021953f4/yjeanrenaud/yj_nearbyglasses
The Nearby Glasses app warns you when smart glasses are nearby by detecting Bluetooth signals from specific manufacturers. It uses company ID codes to identify devices, but may produce false positives due to similar devices.

Goodbye InnerHTML, Hello SetHTML: Stronger XSS Protection in Firefox 148

https://hacks.mozilla.org/wp-content/themes/Hax/img/hacks-meta-image.jpg
Firefox 148 ships the standardized Sanitizer API to prevent XSS attacks. The API sanitizes untrusted HTML before inserting it into the DOM, making the web safer for everyone.

Large-Scale Online Deanonymization with LLMs

https://substackcdn.com/image/fetch/$s_!mY8j!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fccd07a25-d443-4b48-ac20-b10a1e52285a_2002x1292.png
Researchers used LLMs to identify individuals from anonymous online posts with high precision, posing significant privacy concerns. To mitigate this, platforms should restrict data access and LLM providers should implement stronger safety guardrails.

Mercury 2: Fast reasoning LLM powered by diffusion

https://framerusercontent.com/images/RoZiv6hJStkiJmnDuDeCk9S2s.webp?width=512&height=341
Mercury 2 is a fast reasoning language model that generates responses through parallel refinement, producing multiple tokens simultaneously and converging over a few steps. It achieves >5x faster generation with a fundamentally different speed curve, making it suitable for real-time applications.

Hacking an old Kindle to display bus arrival times

https://assets.mariannefeng.com/portfolio/kindle/kindle-finished-product.jpg
The user turned an old Kindle into a live bus feed display by jailbreaking it, installing custom apps, and setting up a server to generate and serve bus data as an image. They created a custom KUAL app to control when the dashboard is displayed and added a feature to exit dashboard mode by pressing the menu button.

How to talk to anyone, and why you should

https://i.guim.co.uk/img/media/806c0c1529c7c258159b7c42834c5d1e4a3d8aec/0_0_4687_3730/master/4687.jpg?width=445&dpr=1&s=none&crop=none
The author reflects on the decline of casual conversations with strangers in public spaces due to social anxiety and overreliance on technology. They argue that this loss of human interaction is a significant issue affecting people of all ages.

Show HN: Steerling-8B, a language model that can explain any token it generates

https://www.guidelabs.ai/images/embedding_decomposition.apng
We are releasing Steerling-8B, a new interpretable model that can trace generated tokens to input context and training data. It achieves competitive performance with significantly less compute and unlocks capabilities like concept control and training data provenance.

Show HN: Moonshine Open-Weights STT models – higher accuracy than WhisperLargev3

https://raw.githubusercontent.com/moonshine-ai/moonshine/main/images/lang-chart.png
Moonshine Voice is a speech-to-text framework that offers fast, private, and accurate live voice interfaces with low latency, supporting multiple languages and platforms. It provides a high-level API for developers to build voice applications with features like transcription, speaker identification, and command recognition, making it easy to integrate voice interfaces into applications.

Steel Bank Common Lisp

SBCL is a high-performance, open-source Common Lisp compiler with a permissive license. It provides an interactive environment with debugging tools and runs on various platforms including Linux and Windows.

Show HN: X86CSS – An x86 CPU emulator written in CSS

https://lyra.horse/x86css/static/thumbnail.png
x86CSS is a CSS-only x86 CPU/emulator that runs native 8086 machine code without JavaScript. It can be built and run using Python scripts and GCC, with a custom I/O interface for user interaction.

Stripe valued at $159B, 2025 annual letter

https://images.stripeassets.com/fzn2n1nzq965/42EFKVMF30ZAldLXtTHYWM/1632284eb36fc3df5aa5a1d115cbd4fb/annualletter2025_newsroom2000x1000.png?w=2000&q=80
Stripe generated $1.9 trillion in total volume, up 34% from 2024, and powers 90% of the Dow Jones Industrial Average and 80% of the Nasdaq 100. Stripe's Revenue suite is on track to hit $1 billion and the company is investing in product development and acquisitions.

Tesla registrations crash 17% in Europe as BEV market surges 14%

https://electrek.co/wp-content/uploads/sites/3/2025/04/Tesla-family-hero.png?w=1600
Tesla's EU registrations dropped 17% in January 2026, while the broader battery-electric vehicle market grew 13.9%. BYD's registrations surged 165% year-over-year, more than doubling Tesla's volume in the region.

OpenAI resets spending expectations, from $1.4T to $600B

https://image.cnbcfm.com/api/v1/image/108267180-17714908241771490821-44105403042-1080pnbcnews.jpg?v=1771490824&w=750&h=422&vtcrop=y
OpenAI plans to spend $600 billion by 2030 and target $280 billion in revenue. The company is finalizing a massive funding round that could total over $100 billion.

Osaka: Kansai Airport proud to have never lost single piece of luggage (2024)

https://japannews.yomiuri.co.jp/wp-content/uploads/2024/12/JW-Kix.jpg
Kansai Airport in Japan has a 30-year record of no lost baggage incidents, earning it the "World's Best Airport for Baggage Delivery" title. The airport uses sensors and manual checks to ensure efficient baggage handling, but faces a staff shortage ahead of the 2025 Osaka-Kansai Expo.

An interactive intro to quadtrees

https://growingswe.com/blog/quadtrees/opengraph-image?16ed498263341a16
A quadtree is a data structure that organizes points in two-dimensional space by dividing the space into smaller regions. It adapts to the data distribution, subdividing dense areas and keeping sparse areas coarse.

IRS Tactics Against Meta Open a New Front in the Corporate Tax Fight

Please enable JS and disable any ad blocker

Shatner is making an album with 35 metal icons

https://cdn.mos.cms.futurecdn.net/ddz75ahfYJmpd62SBPQC2c-840-80.jpg
William Shatner is releasing a metal album with an all-star cast of guitarists, including Zakk Wylde and Chris Poland. The album promises massive guitars, cinematic arrangements, and raw emotion, with Shatner hand-picking 35 metal icons for the project.

Show HN: Emdash – Open-source agentic development environment

https://raw.githubusercontent.com/generalaction/emdash/main/docs/public/media/product.jpeg
Emdash is a tool for parallel development and testing with multiple agents, supporting 21+ CLI providers. It securely connects to remote machines and stores app state locally, with optional GitHub features.

Hugging Face Skills

https://opengraph.githubassets.com/846baf09e6bd1a140786d3aedfc760a9bdd2ddb6b31a277051158e832bde5937/huggingface/skills
Hugging Face Skills are self-contained folders for AI/ML tasks like dataset creation and model training. They are interoperable with major coding agent tools like OpenAI Codex and Google DeepMind's Gemini CLI.

Show HN: enveil – hide your .env secrets from prAIng eyes

https://opengraph.githubassets.com/2ca0c854545dc93fafcc36da28cfbb225fc40344a55e97a1810a3cdf468d4ee8/GreatScott/enveil
enveil is a tool that encrypts secrets in a project directory, preventing AI tools from accessing them. It uses a master password to derive a 256-bit AES key for encryption and decryption.

Looks like it is happening

User speculates about AI-generated papers flooding hep-th submissions, citing rising numbers on arXiv. User asks someone to analyze AI-generated papers and their impact on hep-th submissions.

Iowa farmers are leading the fight for repair

https://valkyrie.cdn.ifixit.com/media/2019/10/22145631/john_deere.jpg
Iowa farmers are pushing for the Right to Repair bill, HSB 751, which aims to give farmers access to diagnostic tools and software needed to repair their equipment. The bill has a good chance of passing in the Iowa House after advancing in the Agriculture Committee with an 18-5 vote.

Cell Service for the Fairly Paranoid

https://cdn.sanity.io/images/9u514z6s/production/d7c261b3976bd7fbd17ca1e0c46177dfb3ae112a-4323x2610.png?w=3000&h=1811&q=100&auto=format
Cape is a privacy-first mobile carrier offering unlimited 5G and 4G data, secure global roaming, and features like Identifier Rotation and Last-Mile Encrypted Texting. It provides a secure alternative to traditional carriers with features like Proton VPN and encrypted voicemails.

Why the KeePass format should be based on SQLite

https://mketab.org/blog/sqlite_kdbx/additional_attributes.png
KeePass's KDBX file format has limitations, including incompatibility with modern security features and large database sizes, which can be addressed by switching to an SQLite-based format. This change would allow for better scalability, security, and user experience, and would facilitate the addition of new features and backward compatibility.

Hegseth gives Anthropic until Friday to back down on AI safeguards

https://media.cnn.com/api/v1/images/stellar/prod/20260224-dario-amodei-pete-hegseth-split.jpg?c=original&q=w_860,c_fill
Defense Secretary Pete Hegseth threatened to terminate Anthropic's $200 million contract and blacklist the AI company if it doesn't comply with demands to lift safeguards on its AI model. Anthropic is refusing to budge on restrictions for AI-controlled weapons and mass domestic surveillance.

Show HN: Hacker Smacker – Spot great (and terrible) HN commenters at a glance

https://hackersmacker.org/images/screenshot3.png
Hacker Smacker is a browser extension for Hacker News that helps identify quality authors and filter out obnoxious commenters. It uses a friend/foe system inspired by Slashdot and is open-source on GitHub.

1Password pricing increasing up to 33% in March

1Password is updating its Family plan pricing starting March 27, 2026, for renewals on or after that date. Current renewals before March 27, 2026, will keep the current pricing until their next renewal.

Stripe reportedly makes offer to acquire PayPal

https://image.cnbcfm.com/api/v1/image/108269085-17719409881771940985-44200112386-1080pnbcnews.jpg?v=1771940987&w=750&h=422&vtcrop=y
PayPal's stock surged 7% after reports of potential acquisition by Stripe, which hit a $159 billion valuation and aims for $1 billion annual revenue. PayPal faces slowing growth and competition, having plummeted 19% this year.

Learnings from 4 months of Image-Video VAE experiments

https://www.linum.ai/vae-reconstruction-vs-generation/high_quality_reconstructions/images/042_original_resized.jpg
Researchers built an Image-Video VAE to compress images and videos into a compact latent space for diffusion models, but found that better compression doesn't always track with VAE stability or downstream generation quality. They discovered that over-optimizing reconstruction quality can hurt the model's chance to disentangle data into a semantically meaningful space, and that aligning the ...

Denver dumps Flock, awards contract to Axon

"Just a little detail that wouldn't sell anything"

https://unsung.aresluna.org/_media/just-a-little-detail-that-wouldnt-sell-anything/ogimage.png
The breathing light, also known as the Sleep Indicator Light, debuted in 1999 iBook G3 and was designed to mimic human breathing for a soothing effect. It was eventually phased out in the 2010s due to design changes.

The quixotic team trying to build a world in a 20-year-old game

https://cdn.arstechnica.net/wp-content/uploads/2026/01/Anvil-640x360.jpg
Thousands of volunteers have collaborated on the Morrowind mod projects Tamriel Rebuilt and Project Tamriel, building a space comparable to a small country. The projects have endured through steady updates, a cohesive team, and improved development tools, with a goal to complete the next task and bring the team closer to the next release.

80386 Protection

https://nand2mario.github.io/posts/2026/80386_protection/80386_labeled_protection.jpg
The 80386 processor implements protection through a dedicated PLA (Programmable Logic Array) called the Test PLA, which evaluates complex privilege rules in parallel, and a hardware state machine for page table walks, allowing for fast and efficient memory protection. The processor's protection architecture is a layered approach, using dedicated hardware for the fast path and microcode for ...

Reading English from 1000 AD

https://lewiscampbell.tech/img/icon512-967261d058848800.png
A post about Old English has gone viral, sparking interest in the language. A non-expert was able to read and understand Old English from 1000 AD with some effort.

Dissecting the CPU-memory relationship in garbage collection (OpenJDK 26)

https://norlinder.nu/assets/img/stopwatch.png
The traditional pause time metric for measuring garbage collection (GC) performance is no longer sufficient due to the increasing complexity of modern GC algorithms and the decoupling of pause time from machine resources. A new Java API and unified logging mechanism in OpenJDK 26 provide a precise and internal accounting of the collector's work, enabling developers to understand the ...

Sprites on the Web

https://www.joshwcomeau.com/images/illustrations/gold-trophy-flames.png
The user explains how to create animations using sprites in CSS, specifically using the steps() function to create discrete animation steps, and discusses the benefits and trade-offs of this approach. The user also shares examples of when to use sprites, such as for dynamic animations that look like sprites, and when not to use them, such as for complex animations that can be handled by modern devices.

Cardiorespiratory fitness is associated with lower anger and anxiety

Russian soldiers tell BBC they saw fellow troops executed on commanders' orders

https://ichef.bbci.co.uk/news/480/cpsprodpb/7a4b/live/90dacc00-0d7c-11f1-8e6b-0f78c649f3c2.png.webp
Russian soldiers have exposed brutal conditions on the front lines in Ukraine, including executions of comrades for refusing orders. They describe "meat storm" missions where troops are sent to their deaths in relentless waves.

Modeling cycles of grift with evolutionary game theory

https://www.oranlooney.com/post/grifters-skeptics-marks_files/lead.jpg
A mathematical model based on Evolutionary Game Theory suggests that grift is cyclical, with periods of high grift followed by high skepticism. The model implies that the current generation of grifters will eventually stop working as people learn to spot their tricks.

We are changing our developer productivity experiment design

https://metr.org/assets/images/uplift-2026-post/uplift_timeline.png
Researchers studied AI's impact on developer productivity, but their experiment had selection effects due to lower pay and AI adoption. They found a likely speedup from AI tools, but their data is weak evidence due to biased developer selection.

Show HN: Linex – A daily challenge: placing pieces on a board that fights back

An addictive line-completion puzzle game. Compete in daily challenges, climb the rankings and prove who’s really the best. Play now!

AI-generated replies are a scourge these days

Something went wrong, but don’t fret — let’s give it another shot. Some privacy related extensions may cause issues on x.com. Please disable them and try again.

Sam Altman Is Losing His Grip on Humanity

https://cdn.theatlantic.com/thumbor/1c1613xpTrAiKaTl5Arsri50blo=/0x0:3710x2087/960x540/media/img/mt/2026/02/2026_02_23_sam_altman/original.jpg
Sam Altman, OpenAI CEO, compared training a human to training a chatbot, saying it takes 20 years of life and energy to become smart. This comparison is alarming as it suggests AI is comparable to humans, potentially justifying treating humans and the planet as collateral damage.

How to Allocate Memory

Optimizing memory allocation can lead to significant speed improvements by rethinking memory usage and using specific data structures like arrays and stacks. Custom allocators can be designed for each data structure, using techniques like reference counting and separate free lists for each class to minimize fragmentation and improve performance.

Sovereignty in a System Prompt

https://pop.rdi.sh/sovereignty-in-a-system-prompt/prompt-extraction.png
Sarvam AI's Indus model has been criticized for its lack of transparency and potential ideological bias, with a system prompt instructing it to prioritize national pride and dismiss international scrutiny. The model's training data and pipeline control are also questioned, with some suggesting it may have been heavily influenced by Nvidia.

Manjaro website off-line again due to lapsed certificate

https://distrowatch.com/images/k/txd.png
Manjaro website is offline due to expired HTTPS certificate. This is the third time the site has gone down due to certificate renewal issues.

PCB Tracer

A free online tool allows PCB reverse engineering with virtual X-Ray view and annotation of components and signal paths. It features AI-powered component identification and no installation is required.

Fed's Cook says AI triggering big changes, sees possible unemployment rise

Please enable JS and disable any ad blocker

Distributed Systems for Fun and Profit

https://book.mixu.net/distsys/images/barroso_holzle.png
Distributed systems are about solving problems that can't be solved on a single computer by using multiple computers, and they involve dealing with distance, time, and consistency models. The key concepts in distributed systems include scalability, availability, performance, latency, and fault tolerance, which are achieved through techniques such as partitioning and replication, and ...

Intel XeSS 3: expanded support for Core Ultra/Core Ultra 2 and Arc A, B series

You don't have permission to access "http://www.intel.com/content/www/us/en/download/785597/intel-arc-graphics-windows.html" on this server.

xAI and Pentagon reach deal to use Grok in classified systems

xAI's Grok model will be used in classified military systems after signing an agreement with the Pentagon, replacing Anthropic's Claude model. The Pentagon is threatening to sever ties with Anthropic over safeguards and is also in talks with Google and OpenAI for their models.

I think WebRTC is better than SSH-ing for connecting to Mac terminal from iPhone

https://macky.dev/app_icon_trans.png
Signaling validates identity tokens for access but a Master Password protects the terminal, preventing access even if your account is compromised. “I can talk with my BFF Claude on the toilet instead of doom scrolling.”

The Hydrogen Truck Problem Isn't the Truck

Hydrogen fuel cell trucks work commercially but face significant efficiency and infrastructure gaps compared to battery electric vehicles. They require 2.5-3x more renewable electricity to deliver the same work.

Meta problem with URPF our bundle in Boca raton

https://metafixthis.com/wp-content/uploads/2026/01/ext-custom-logo-1769900600309.webp
Meta's clusters in Boca Raton and Miami are experiencing packet loss and probable BGP or IGP issues. This is causing data corruption and slow performance for MNA content delivery and direct content consumption in Latin America.

The history of knocking on wood

https://substackcdn.com/image/fetch/$s_!UVR5!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ff288319a-ddeb-4403-a917-0321aaeab2b0_2606x1844.png
The author reflects on the history of gesture, exploring how it is shaped by both nature and nurture, and how it differs across cultures and time. This inquiry into the history of gesture has implications for AI alignment, as machines lack embodied experience and cannot replicate human norms rooted in physicality.

Package Managers à la Carte: a formal model of dependency resolution

https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png
Researchers propose Package Calculus, a unified formalism for dependency resolution across diverse package managers, enabling translation and resolution across ecosystems. This formalism models real-world package managers' diversity and addresses fragmentation in multilingual projects.

Show HN: Scheme-langserver – Digest incomplete code with static analysis

https://raw.githubusercontent.com/ufo5260987423/scheme-langserver/main/doc/figure/auto-completion.png
WANG, Z. has developed scheme-langserver, a language server protocol implementation for Scheme, providing completion, definition, and type inference. The project aims to improve the Scheme programming environment and is open-source, with ongoing development and testing, and seeking donations and funding.

NIST Seeking Public Comment on AI Agent Security (Deadline: March 9, 2026)

NIST requests information on secure AI agent system development and deployment to advance US economic competitiveness and public safety. The RFI seeks insights on security threats, risks, and vulnerabilities, as well as technical controls and practices to ensure AI agent system security.

Why High FOV Sucks – Fixing It with Panini Projection

Show HN: Beehive – Multi-Workspace Agent Orchestrator

https://storozhenko98.github.io/beehive/logo.png
~ git status On branch feat/oauth Changes not staged: modified: src/auth.ts modified: src/middleware.ts ~ npm test PASS src/auth.test.ts (12 tests) PASS src/middleware.test.ts (8 tests) ~ ┌ │ │ Welcome back! │ │ Opus 4.6 │ ~/feature-auth │ └ ❯ add OAuth2 flow to src/auth.ts ❯ Full git clones on any branch, each in its own directory. No interference between experiments.

Eschewing Zshell for Emacs Shell (2014)

The user has found that Emacs' eshell is a better fit for their workflow than traditional shells like Zshell and Bash, due to its integration with Emacs and Lisp-like syntax. They have customized eshell with their own predicates and modifiers, including one to filter files based on their internal Org #+tags entry, and have also made improvements to eshell's behavior, such as farming out ...

Software 3.1? – AI Functions

https://blog.mikegchambers.com/assets/images/ai-functions-2.png
Andrej Karpathy proposes Software 3.1, a point release of Software 3.0, where LLMs generate and execute code, and machine verifies it at runtime. AI Functions, a new project, enables this by using post-conditions to validate output and trigger retries, making it a more robust and self-correcting system.

Decimal-Java is a library to convert java.math.BigDecimal to and from IEEE-754r

https://opengraph.githubassets.com/fc04e240ce01bc211564e587309606164fc0414a2875829f29744320c4ae33ba/FirebirdSQL/decimal-java
Decimal-java is a library to convert Java BigDecimal to IEEE-754r decimal byte representations. It supports Decimal32, Decimal64, and Decimal128 types with optional overflow handling and rounding.

Show HN: SNKV – SQLite's B-tree as a key-value store (C/C++ and Python bindings)

https://raw.githubusercontent.com/hash-anu/snkv/master/demo.gif
SNKV is a lightweight key-value store built on SQLite's B-Tree storage engine without SQL. It provides a clean KV API with direct access to the storage core, bypassing SQL parsing and query planning.

Show HN: Tag Promptless on any GitHub PR/Issue to get updated user-facing docs

Prithvi and co-founder Frances launched Promptless, an AI teammate that drafts doc updates based on workflows. Promptless 1.0 addresses learnings from the last year with features like automatic screenshot updates and improved writing quality.

NPM install is stealing your passwords – I built a tool to catch it

https://westbayberry.com/static/img/clean.png
This tool provides visibility into dependency behavior in CI pipelines, flagging suspicious packages for review and auto-approving the rest. It uses 26 behavioral detectors to analyze package behavior and prevent zero-day attacks, with a self-hosted option for enterprise environments.

How Long Is the Coast of Britain? (1967)

System detected unusual traffic from your network, requires reCAPTCHA completion to verify human activity. Please complete reCAPTCHA to resolve the issue.

The Engine Behind the Hype

https://www.onuruzunismail.com/assets/images/blog/the-engine-behind-the-hype/red-mass-teal-engine.webp
The user discusses their experience with various AI coding tools, including Claude Code, and how they discovered Pi, a custom coding agent built by Mario Zechner, which offers a more efficient and context-aware experience. Pi's simplicity and extension system impressed the user, who believes it's a more effective tool than popular alternatives like Claude Code and OpenCode.

Proposal: Add "AI generated" as a flag reason

The user suggests implementing a flag for LLM-generated content to replace current immune responses of adding a vibecoding tag or flagging as spam. This would help identify and remove low-value AI-generated content, improving the site's quality and user experience.

The Weird OS Built Around a Database [video]

Show HN: Quantifying opportunity cost with a deliberately "simple" web app

https://shouldhavebought.com/images/og-main.png
- LIMITED TO 7 ACTIVE SLOTS - OLDER SLOTS CAN BE DISPLACED - MINIMUM PRESENCE: 7 DAYS - WEEKLY PRESENCE: $500–$1,000 - MONTHLY PRESENCE: $2,000–$3,000 - LIFETIME SLOT: NEGOTIATED (1 ONLY)

The US Had a Big Battery Boom Last Year

https://media.wired.com/photos/6998701dc31375ebca5a4cab/3:2/w_2560%2Cc_limit/GettyImages-2191227367.jpg
The US added a record-breaking 57 gigawatt hours of energy storage in 2025, enough to power over 5 million homes. The growth is expected to continue with a predicted 21 percent increase in 2026.

Show HN: Recursively apply patterns for pathfinding

Scientists crack the case of "screeching" Scotch tape

https://cdn.arstechnica.net/wp-content/uploads/2026/02/scotch1-640x548.jpg
Researchers discovered that the screeching sound from peeling Scotch tape is caused by shock waves from micro-cracks traveling at supersonic speeds. The sound is generated by a train of weak shocks when the transverse cracks reach the edge of the tape.

Dream Recorder AI – a portal to your subconscious

The Righteous EV Owners Who Won't Let Their Broken Cars Die

https://media.wired.com/photos/699892bd88a7d3ff8b60ed09/master/w_2560%2Cc_limit/ERIC_RUBY_20262D0975.jpg
Svein Hodne, a Fisker Ocean owner, was stranded in his car after it lost power and locked him inside, prompting him to join the Fisker Owners Association, a volunteer-run group that took control of the car's maintenance and repair. The association's efforts led to a global movement for the "right to repair," challenging the dominance of tech companies and pushing for greater access to car ...

I turned off ChatGPT's memory

https://d24ovhgu8s7341.cloudfront.net/uploads/editor/advertisements/938/optimized_6c494eba-e388-4bf1-925e-0910ddd7275c.png
Mike Taylor argues that ChatGPT's memory feature has more disadvantages than advantages, causing "context rot" and biased results. He suggests turning off memory for unbiased results, citing his own experiences with custom instructions and personalized recommendations.

Even the Mars Rover Uses Zip Ties (2021)

https://www.thedrive.com/wp-content/uploads/images-by-url-td/content/2021/02/Mars-Rover-2.jpg?quality=85&w=1920
NASA's Perseverance rover uses Ty-Rap cable ties made by ABB to secure wires and equipment. These durable ties can withstand extreme heat and radiation, also used in previous space expeditions.

Mercury 2: Diffusion Reasoning Model

https://framerusercontent.com/images/RoZiv6hJStkiJmnDuDeCk9S2s.webp?width=512&height=341
Mercury 2 is a fast reasoning language model that generates responses through parallel refinement, producing multiple tokens simultaneously and converging over a few steps. It achieves >5x faster generation with a fundamentally different speed curve, making it suitable for real-time applications.

Psychology suggests making a shopping list is a sign of sharper thinking

https://img.etimg.com/thumb/msid-127796077,width-300,height-225,imgsize-104083,resizemode-75/grocery-list-contemplation.jpg
Writing a shopping list by hand requires cognitive skills like planning, memory, and self-control, which are linked to intelligence and long-term decision-making. This habit reflects traits associated with intelligence, such as foresight, organization, and efficient mental processing.

UPP: Universal Predicate Pushdown to Smart Storage

Continue local sessions from any device with Remote Control

https://claude-code.mintlify.app/_next/image?url=%2F_mintlify%2Fapi%2Fog%3Fdivision%3DPlatforms%2Band%2Bintegrations%26appearance%3Dsystem%26title%3DContinue%2Blocal%2Bsessions%2Bfrom%2Bany%2Bdevice%2Bwith%2BRemote%2BControl%26description%3DContinue%2Ba%2Blocal%2BClaude%2BCode%2Bsession%2Bfrom%2Byour%2Bphone%252C%2Btablet%252C%2Bor%2Bany%2Bbrowser%2Busing%2BRemote%2BControl.%2BWorks%2Bwith%2Bclaude.ai%252Fcode%2Band%2Bthe%2BClaude%2Bmobile%2Bapp.%26logoLight%3Dhttps%253A%252F%252Fmintcdn.com%252Fclaude-code%252FTBPmHzr19mDCuhZi%252Flogo%252Flight.svg%253Ffit%253Dmax%2526auto%253Dformat%2526n%253DTBPmHzr19mDCuhZi%2526q%253D85%2526s%253Dd535f2e20f53cd911acc59ad1b64b2e0%26logoDark%3Dhttps%253A%252F%252Fmintcdn.com%252Fclaude-code%252FTBPmHzr19mDCuhZi%252Flogo%252Fdark.svg%253Ffit%253Dmax%2526auto%253Dformat%2526n%253DTBPmHzr19mDCuhZi%2526q%253D85%2526s%253D28e49a2ffe69101f4aae9bfa70b393d0%26primaryColor%3D%25230E0E0E%26lightColor%3D%2523D4A27F%26darkColor%3D%25230E0E0E%26backgroundLight%3D%2523FDFDF7%26backgroundDark%3D%252309090B&w=1200&q=100
Remote Control allows you to start a task on your machine and pick it up from another device. It runs locally and keeps your environment and project configuration available.

Show HN: Out Plane – A PaaS I built solo from Istanbul in 3 months

https://outplane.com/api/og?title=Developer%20Infrastructure%20Platform&category=Out%20Plane
Out Plane automates deployment and scaling with a single GitHub push, reducing deploy times to under 2 minutes. It offers transparent pricing, automatic scaling, and 99.99% uptime SLA with real-time metrics and logs.