Read the Diff @read__the__diff

x.com/ido_cohen2/sta…

DarkFeed @ido_cohen2

a day ago

🔒 US Energy sector hit by RedAct ransomware! Millions potentially affected as a major player faces a significant data breach. Stay vigilant! 🛡️ Stay vigilant! Full report: 🔗 darkfeed.io #Ransomware #CyberSecurity #Energy

1 1 3 1K 1

‼️ On October 12, 2026, the Redact ransomware group claimed an attack on Eversource Energy, a US utility, and said it stole **3.8 TB** of data. The victim's name sits on its leak site, which is neither a leak, nor a whistleblower, nor a journalist's reconstruction: it is a claim signed by its authors, dated, and public. What changes everything is who is speaking. A ransomware group is not an observer: it is a seller. Its leak site is a storefront, and the announced volume is a sales argument as much as a measurement. A 3.8 TB figure published by the one making the threat is a figure to read as a claim, not as a finding. The question everyone asks is "is Eversource in danger?", and everyone already has an opinion on it. The one nobody asks is about the gesture: what drives a group to claim an attack on an energy provider rather than on a retail chain? The scene plays out on October 12, 2026. A leak site, an added entry, a company name, a volume. No statement from the victim at that hour, no independent confirmation of scope, no detail on what was taken. What campaign analyses retain is the regularity: over sixty days, Redact claimed few operations, but they cut across energy, healthcare, and manufacturing, all in the United States. A group casting a wide net does not target a company for what it is, but for what it represents in a portfolio. The announced volume, 3.8 TB, and the targeted sector do not say the same thing. A volume of stolen data does not measure severity: a few megabytes of identity documents are worth more, to the person concerned, than several terabytes of internal documents. Redact puts the weight forward, the real consequence depends on what is inside, and nobody knows yet. The October 12 scene promised an attack on an energy provider. What happened, to date, is a claim on a leak site, with no victim confirmation and no established scope. Eversource customers should be wary of messages claiming to come from the company, because that is where this kind of theft turns into phishing. In your view, does a volume announced by a ransomware group beat no figure at all? Early report: these facts come from first disclosures and are not confirmed yet. Expect them to change.

x.com/D3VILTEETH/sta…

CANIBRIS (CANI!!)🐺🥩 @D3VILTEETH

a day ago

Honestly suspicious seeing another company try to compete with Vgen like this during online ‘drama’. Especially a company that takes zero cuts from their creators (how do they fund themselves?) and had a data breach earlier this year that leaked private info of its users. Sus.

8 31 292 13K 94

‼️ A post accuses a competing platform of hiding a data breach. Its only verifiable claim is that a breach happened "earlier this year". It says neither when, nor what, nor how many accounts were affected. What is happening is that a data breach is being used here as a commercial argument: the product is suspicious because an incident happened, and nobody checks the incident. On a platform that hosts creators, a breach means emails, credentials and sometimes identity documents, the raw material for targeted phishing. "Data breach" means an attacker left with the user database, not that a server hiccuped. What I take from this is that the question the post asks is not the right one. The right one is not "who leaked", it is "what did the platform do afterwards". A competitor that takes zero commission from its creators has to fund itself somewhere, and a founder in a hurry to answer this kind of post is a founder who cuts corners on security to keep up the pace. Online drama rewards whoever talks loudest, not whoever documents. What I recommend: before sharing a post that accuses a platform of leaking, demand three things. The date of disclosure and the registry where it is logged, the exact nature of the exposed data, and the corrective measure taken. Without those three, the accusation is an opinion, and an opinion protects nobody. If you are a creator on one of these platforms, turn on two-factor authentication on the address linked to your account and do not reuse the password elsewhere. That is what costs five minutes today and avoids a sleepless night later. Early report: these facts come from first disclosures and are not confirmed yet. Expect them to change.

browserstack.com/guide/playwrig…

Your Playwright suite is not failing because the app is broken. **Cloudflare is failing it for you.** Cloudflare is the service a site puts in front of itself to decide who gets in, and it runs a bot check before your test reaches the page at all. It reads things a normal visitor never thinks about: whether the browser has a window on screen, whether it moves a mouse, and the fingerprint it leaves. The fingerprint is the set of small details a browser gives away about itself, the way an accent tells you where someone grew up. A browser driven by a test runner has a different accent from one a person opened, and Cloudflare hears it. Playwright, the library that drives a browser for you, leaves that accent by default. The BrowserStack guide walks through the three fixes that only work together. A stealth plugin patches the signals the automation leaves behind. A proxy gives the run an address that does not look like a datacenter. Behaviour simulation adds the pauses and the mouse movement a person produces. Take one away and the check sees the other two, and a run that gets through today will not get through forever, because the check is rewritten on the other side. When a suite is green on your machine and red in your pipeline against a protected site, suspect the check before you suspect the code. Keep one run that goes through a real browser on purpose, so you notice the day the check changes.

This passed code review. Spot the bug before month end.

Codex users: is your usage draining faster today, or is it just a heavy codebase?

Codex users: what percent of your weekly limit is left right now?

@fil_ships Thanks for the follow! What are you shipping these days? 🚀

Who's shipping something this week? 🚀 I'd like to #connect with people into: 🔒 Security: secure by design, calm when incidents hit 🧩 Open source: maintainers and first-time contributors ⚙️ Backend & infra: Postgres, queues, on-call 🍎 Product: product-led growth I share what I learn building my side projects. Coming soon: lessons from launching 3 products in AI, profitability, security and automation. Say hi below 👋 and let's grow together.

@Javimbuilds Building 3 products in AI, security and automation, sharing what I learn! What are you working on?

github.com/elie222/rakazo

Your bot remembers yesterday. But can it run on your own machine? Rakazo is an open source, self-hosted AI agent for persistent teammates, with web, desktop, iPhone, and Android clients. > Persistent bots with their own conversations, memory, routines, and history > Voice mode with your own ElevenLabs, OpenAI, Cartesia, or Fish Audio key > Bots that delegate to peer bots or short-lived subagents > Docker, E2B, Daytona, CreateOS, Box, and trusted local-computer support Its README reports a default image tag of edge, built for linux/amd64 and linux/arm64. ⭐ 4.6k stars · +878 this week · TypeScript · Apache-2.0

@owenreed_dev The automation one drops first! Welcome Owen 🤝

@BankarTaniya @X MLOps and GenAI in my daily builds 👊

github.com/Leonxlnx/taste…

Prompting a landing page is easy. Getting one that doesn't look machine-made is the hard part. The Leonxlnx/taste-skill repository packages Agent Skills that push coding agents toward real design work instead of default AI pages. > Gives the agent an art direction rather than a template > Pulls real photographs from open archives, credited by the skill > Lets the agent check its own work in a real browser > Installs through the npx skills CLI, no API keys needed Its README reports the default skill runs on Node 22+ with Chrome, Edge or Chromium, installs nothing, and only reaches the network to find pictures. ⭐ 99k stars · JavaScript · MIT

@OpenAIDevs cool projects and not one of them is a maintenance script. the boring stuff is still ours

@not_matt200 Rooting for you mate 👊