🛠️ Hacker News Tools

Show HN: Mail Memories – A desktop app to rescue photos from Gmail Hey HN, I’m the creator of Mail Memories. Like many of you, I&#x27;ve had my Gmail address for more than 20 years. A few years ago, I got curious and wanted to see what photos were buried deep in my account. I ended up finding lots of &quot;lost&quot; pictures of old friends, family members, and a ridiculous number of vintage memes.<p>I originally built and launched this as a SaaS, but even with code and policies in place that kept users&#x27; photos private, I figured everyone would feel more comfortable with a desktop app.<p>So, I threw out the server architecture and completely rewrote it as a 100% local desktop app for Mac and Windows.<p>How it works now: The app connects directly to Google&#x27;s server from your computer, processes everything entirely on your system, and saves photos straight to your hard drive.<p>You can download your 50 oldest photos for free (no credit card required) just to see what&#x27;s in there. If you want to download all the pictures in your account, it&#x27;s a one-time payment of $29. No subscriptions.<p>If you have an old, pre-2010 Gmail account, definitely give it a spin. You&#x27;ll be surprised at what you find deep in your archive.<p>I&#x27;d love to hear your feedback on the layout, scanning performance, or anything else.<p>TL;DR: I turned my SaaS into a local desktop app (Mac&#x2F;Windows) that recovers decades of forgotten photos from your Gmail. 100% local, no cloud, no subscriptions, no AI.

Show HN: ZeroFS – A log-structured filesystem for S3

Show HN: I built an open-source alternative to Claude Cowork Hey HN,<p>A few months ago, I tried to automate some of my work with the popular AI agent OpenClaw, and then I quickly realized how difficult it is to get it to work with APIs and third-party services securely, which is essential for a lot of work-related tasks.<p>Then I realized OpenClaw is more of a personal assistant and it was not designed to get actual work done as a coworker. So I started to build Valmis, an alternative to OpenClaw that works with more than 100 apps and services, with security being the priority.<p>Valmis addresses the security issue by designing a proxy system: dockerized agent runtime can only request the host machine to make API requests by providing the relevant credential ID. The host then makes the actual request and returns the JSON data to the agent runtime. With this design, you can even turn off the internet access of the agent container while making it work.<p>Our proxy system now supports 100+ business and productivity integrations, including all Google Workspace apps, Slack, Notion, HubSpot, Salesforce, and Figma.<p>One of the coolest features of Valmis is the automated workflow. You can automate multi-step workflows using our workflow builder. Each workflow can be triggered by cron, webhooks, app events, and it supports conditions and loops.<p>I&#x27;d be happy to answer any questions in the comment section.

This blog is written in en-GB

Beyond Git: Real-Time Version Control for Godot – Lilith Duncan – GodotCon 2026 [video]

Is the iPhone birth control? Causal evidence from AT&T's 2007-2011 monopoly [pdf]

Show HN: Kube-insight – retained Kubernetes evidence for incident investigations

The primary purpose of code review is to find code that will be hard to maintain

Show HN: I measured the half-life of 41,301 Show HN launches. It's 7 hours I scraped every Show HN from the last 12 months (41,301 posts) plus the full comment tree of every launch with 10+ comments, ~100k comment timestamps, all from the Algolia HN API.<p>The median launch gets 2 points and 0 comments. For launches that do get traction, half the comments they&#x27;ll ever get arrive within 7.2 hours and 90% within 26, and the top decile decays on the same clock as everyone else.<p>Vote timestamps aren&#x27;t public, so comment timing is the attention proxy; caveats are in the post. Everything reproduces from the repo with one command (<a href="https:&#x2F;&#x2F;github.com&#x2F;jonnonz1&#x2F;hn-attention-cliff" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;jonnonz1&#x2F;hn-attention-cliff</a>), and every number in the post maps to a named function. Keen to hear where the methodology falls short

PeerTube is a free, decentralized and federated video platform

The great blogging collapse: What happened to 100 successful blogs?

Evaluation order and nontermination in query languages

The Complete Homemade Juggling Beanbag Guide

He sent a harsh email to ICE's top official. Federal agents tracked him down

Kimi K2.7 Code is generally available in GitHub Copilot

Show HN: Margarita - Programming language for Agents using Markdown-ish syntax On my list of build it from scratch has always been to build a programming language. So with the help of AI I was able to get it done! Why did I build it? At work I&#x27;ve seen two major problems with our ai workflows&#x2F; skills libraries. There is a lack of determinism when your whole workflow is a markdown file of 100 steps, and markdown skill libraries lack composability. Meaning we violate things like DRY in the all the md files in the skills library.<p>I built Margarita to allow for markdown and logical operators to exist together, which means you can bring in determinism through code structures when it makes sense, and fall back to llm dynamic code when that makes sense. As an added bonus allows for composable prompts ala React which solve my other gripe with skills libraries being a mash of text everywhere.<p>Overall I&#x27;ve been getting pretty luke warm responses from Reddit, so I&#x27;ll probably just shelve it, but it was a blast to make. Got to build code agents for pretty much every llm provider and built my own harness. I would recommend doing that it&#x27;s a great learning experience.<p><a href="https:&#x2F;&#x2F;www.margarita.run" rel="nofollow">https:&#x2F;&#x2F;www.margarita.run</a> <a href="https:&#x2F;&#x2F;github.com&#x2F;Banyango&#x2F;margarita" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;Banyango&#x2F;margarita</a>

Herdr: One terminal to rule them all

Show HN: Meow – The 4th and final JavaScript runtime and toolchain Yes, I said final. Fight me in the comments.<p>But seriously: there are basically exactly 3 serious JS runtimes in the world. (Now 4... or 1, idk).<p>Why? Because the modern JS ecosystem is a fragmented, bloated nightmare. Node, pnpm, tsc, eslint, prettier, vite. Each spins up its own V8 instance, parses your codebase from scratch, and leaves a graveyard of .json configs in your root directory.<p>meow squishes it all into one cute 82MB binary. It’s a drop-in Node replacement, ultra-fast package manager, formatter, test runner, linter, checker and bundler.<p>I didn&#x27;t write a JS engine from scratch, I&#x27;m not that depressed. meow uses V8 for stability and gets Node drop-in compat via deno_node (thanks Deno team).<p>The secret sauce: meow parses your codebase exactly once into an Oxc AST in memory. That single graph natively feeds the runtime, linter, formatter, typechecker, and bundler. No redundant allocations. Just the fastest JavaScript tooling available, completely unified.<p>0-Byte Duplicated node_modules (Without Symlink Hell) pnpm is cool, but its symlinks notoriously break framework resolvers. meow downloads to a global cache once, then projects packages using OS-level copy-on-write (macOS APFS clonefile) or parallel hardlinking (Linux&#x2F;Windows). You get flat, standard node_modules consuming exactly 0 extra bytes on your SSD.<p>7ms Warm Installs Offloading SHA-512 cryptographic verification to background OS threads means cold installs take ~250ms, warm installs take 7-15ms. And yes, it does it that fast without skipping verification.<p>It actually runs Next.js &amp; SvelteKit Fast runtimes are a fun parlor trick until they fail to build your app. meow natively boots Next.js 15, Astro, and SvelteKit out of the box with full CommonJS &amp; Node built-in support.<p>Sandboxed npx by default Running npx create-next-app gives a random script full host access. We shouldn&#x27;t be letting Shai Hulud run around so freely. meow x (our ephemeral runner) is sandboxed by default. Network is denied, writes are confined to the cwd, and the clock&#x2F;entropy are frozen. Pass --trust to let it do sketchy stuff.<p>The UX meow looksmaxxes hard. It looks like a little girl vomited in a TTY. But it intelligently detects non-interactive environments and degrades to clean stdout so you can sneak it into your org without breaking CI&#x2F;CD.<p>No cap, it benchmaxxes pretty good too. Like if it were an LLM it&#x27;d be mythos class for sure. Benchmarks below.<p>Install (macOS &#x2F; Linux): curl -fsSL <a href="https:&#x2F;&#x2F;meow.style&#x2F;install" rel="nofollow">https:&#x2F;&#x2F;meow.style&#x2F;install</a> | sh Web: <a href="https:&#x2F;&#x2F;meow.style" rel="nofollow">https:&#x2F;&#x2F;meow.style</a> Docs: <a href="https:&#x2F;&#x2F;docs.meow.style" rel="nofollow">https:&#x2F;&#x2F;docs.meow.style</a> GitHub: <a href="https:&#x2F;&#x2F;github.com&#x2F;0xchasercat&#x2F;meow" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;0xchasercat&#x2F;meow</a><p>Okay, who wants to fight me? Come hit me with the &quot;but it don&#x27;t work with xxxx!1!!!&quot;. (But really, if you do encounter an issue, GH issues are welcome).<p>Disclaimer: meow is still super early stage. You will hit edge cases, but for common apps it works perfectly.<p>Benchmarks (<a href="https:&#x2F;&#x2F;github.com&#x2F;0xchasercat&#x2F;fluffybench" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;0xchasercat&#x2F;fluffybench</a>) Package Manager (Large workspace, Cold &#x2F; Dedup): Meow: 248ms &#x2F; 725ms Bun: 2.7s &#x2F; 3.5s NPM: 36.7s &#x2F; 55.9s Deno: 46.7s &#x2F; 55.0s Runtime (HTTP req&#x2F;s | Regex | Cold Start): Meow: 43.7k req&#x2F;s | 89ms | 23ms Deno: 42.0k req&#x2F;s | 90ms | 24ms Node: 40.3k req&#x2F;s | 99ms | 38ms Bun: 36.2k req&#x2F;s | 107ms | 16ms

Android Developer Verification: Threat masquerading as Protection

Artificial Adventures