application Archives

AppWizard

May 2, 2026

Google releases experimental ‘COSMO’ AI assistant app on Play Store

Google released an experimental AI assistant application called COSMO for Android devices, which has since been removed from the Play Store due to an accidental launch. The app, which is 1.13 GB in size, was designed to enhance daily productivity by organizing schedules and answering complex queries. COSMO features include a list tracker, document writer, calendar event suggester, browser agent, timer suggestions, deep research capabilities, quick photo lookup, web search functionality, jargon definitions, insights provision, context understanding for people and events, recall assistance, and conversation summarization. It offers three settings: Hybrid, PI Only, and Nano Only, with the latter two indicating different processing methods. The app also includes voice match and screen access features but is not intended for the general consumer market.

Winsage

May 1, 2026

Microsoft releases first big update after Nadella’s vow to ‘win back fans’

Microsoft released a non-security update on April 30 aimed at enhancing Windows stability and performance, particularly improving Windows Explorer functionality. The update improves the reliability of explorer.exe processes, enhances sign-in processes, interactions with taskbar menus, and unpinning items from Quick Access. It optimizes startup application performance and storage performance, raising the maximum FAT32 format size limit from 32GB to 2TB. Additional reliability improvements were made to Windows Hello, the Microsoft Store, and the taskbar system tray. The update also introduces Enterprise State Roaming (ESR) for device transitions, policy-based removal of preinstalled Microsoft applications, and enhanced security for batch files.

AppWizard

May 1, 2026

Google just dropped a new ‘experimental AI assistant’ app exclusively for Android

Google has introduced an experimental AI assistant application named COSMO to the Play Store, identified by the package name com.google.research.air.cosmo. COSMO aims to enhance daily tasks using artificial intelligence, similar to Google’s Gemini app. The app can operate using a local Gemini Nano model or connect to a remote “PI” server, although some features are not fully operational. The user experience is currently rough, with less polish compared to the full Gemini app, and access is limited across various regions, including on Pixel devices.

Winsage

May 1, 2026

Microsoft brings a Linux-style window trick to Windows 11 via PowerToys, and I can’t go back

Microsoft's PowerToys version 0.99 introduces two new tools: Grab And Move and Power Display, along with enhancements to the Command Palette Dock. Grab And Move allows users to reposition application windows by holding the Alt key and dragging anywhere on the window, facilitating easier window management. Users can customize the activation key and toggle the feature off during gaming sessions. There are also options to exclude specific applications from using this feature. Power Display enables users to manage monitor settings directly from the system tray, including brightness, contrast, volume, rotation, and color temperature adjustments. It supports multi-monitor setups with DDC/CI standards and includes a Profiles option for saving specific monitor settings. The Command Palette Dock now features a compact mode and customization options for pinned commands, improving user accessibility and screen space utilization.

Tech Optimizer

May 1, 2026

Backstage Ditches Postgres for Databricks Lakebase

The divide between operational and analytical databases is diminishing, as evidenced by Databricks integrating Spotify's Backstage with its Lakebase, transitioning from PostgreSQL to a unified data platform. Lakebase now offers a serverless PostgreSQL interface within Databricks, allowing Backstage to operate without awareness of the underlying changes. The integration involved updating configurations and addressing authentication challenges using OAuth JWTs. The integration transforms the database development lifecycle by enabling near-instant database branching through Lakebase's copy-on-write architecture, allowing developers to focus on testing changes rather than safety. Branching creates a pointer to existing data, making the operation instantaneous. Point-in-Time Recovery (PITR) allows for quick restoration of data, demonstrated by a recovery process that took less than four seconds. The goal is to integrate database branching into developer workflows seamlessly, making it automatic. This change could enhance developer productivity by allowing live data testing from the start, eliminating the need for mock objects, and resolving common issues related to staging environments. The traditional constraints of slow and costly database copies will become obsolete, prompting teams to reconsider how much time they spend on workarounds for now-nonexistent constraints.

Winsage

May 1, 2026

Windows 11 KB5083631 out with Xbox mode, faster performance (direct download links .msu)

Windows 11 update KB5083631 introduces enhancements for user experience on PCs, available for download via Windows Update or direct links. The update is identified as “2026-04 Preview Update (KB5083631) (26200.8328)” and updates the OS to Build 26200.8328 for version 25H2 or Build 26100.8328 for version 24H2. Direct download links for Windows 11 KB5083631 include: - Build 26200.8328, 4802.0 MB, Windows 11 25H2, x64-based - Build 26200.8328, 4802.0 MB, Windows 11 25H2, arm64-based - Build 26100.8328, 4449.0 MB, Windows 11 24H2, x64-based - Build 26100.8328, 4449.0 MB, Windows 11 24H2, arm64-based Key features of the update include: 1. Introduction of Xbox Game Mode to standard PCs, allowing users to access a gaming-centric interface with an Xbox dashboard. The feature is gradually rolling out, and users can enable it manually using ‘ViveTool’. 2. Improvements to File Explorer, including faster performance and consistent folder views, ensuring a reliable experience regardless of access method. Users can download this optional update without major issues reported.

AppWizard

April 30, 2026

LofyStealer Targets Minecraft Players via Node.js Loader and Browser Injection

Minecraft players are being targeted by a deceptive hacking tool called “Slinky,” which is actually an infostealer known as LofyStealer linked to the Brazilian cybercrime group LofyGang. The malware uses a Node.js-based loader and a C++ payload to extract sensitive browser data, disguising itself as a Minecraft hack. It primarily targets younger players who may unknowingly execute it. The malware operates through a two-stage architecture, with a 53.5 MB loader binary that injects a smaller 1.4 MB payload into browser processes, bypassing detection methods. It collects sensitive information such as cookies, passwords, and credit card details, compressing and encoding the data for exfiltration. The command-and-control (C2) infrastructure is hosted in Brazil, and the malware's design reflects advanced compilation techniques. The campaign is attributed to LofyGang, which has evolved into a more professional entity, posing severe risks to players who download cheats and cracked tools. Indicators of compromise include a specific C2 IP address and associated endpoints.

AppWizard

April 30, 2026

LofyStealer Uses Node.js Loader Against Minecraft Gamers

Cybersecurity threat hunters have discovered an active infostealer campaign targeting the gaming community, involving malware called LofyStealer (or GrabBot) that disguises itself as a Minecraft hack named “Slinky.” The attackers use the official game icon to trick young gamers into executing the malware. The Brazilian cybercrime group LofyGang has enhanced its technical capabilities, utilizing a sophisticated two-stage modular architecture. The initial stage features a 53.5 MB loader file named load.exe, which is a Node.js runtime environment that obscures malicious signatures. The loader connects to the attacker’s server and decrypts a 1.4 MB C++ payload, chromelevator.exe, which targets eight web browsers to extract sensitive information like cookies and passwords. The stolen data is compressed, encrypted, and sent to the attacker’s server. LofyGang has evolved into a Malware-as-a-Service platform, offering a web panel for operators to monitor victims and generate custom executables. The campaign highlights the increasing threats to the gaming community, with advanced evasion techniques being employed by cybercriminals. Security professionals are advised to monitor network traffic and conduct audits for suspicious activities.

AppWizard

April 30, 2026

Google AI is creeping into every app on your phone. Here is how to disable it

Google has integrated its Gemini AI into various applications, including Gmail, Google Drive, Messages, and Google Photos. There is no universal switch to disable AI across Android devices due to varying manufacturer implementations. Users can selectively disable AI features in specific apps: - Google Messages: Disable Gemini by going to Messages settings and turning off the Gemini option. - Gmail: Turn off Smart features in Google Workspace and other Google products through the Gmail settings. - Google Photos: Disable Gemini features in Photos settings under Preferences. - Phone app: Disable AI features like Call Screen in the Phone app settings under Call Assist. - Pixel Screenshots: Disable the AI feature for searching screenshots in the Pixel Screenshots app settings. Disabling the Gemini app itself will remove its integrations and deactivate AI features on the device.

Tech Optimizer

April 30, 2026

Inside the Architecture of an MCP Server for PostgreSQL

The Model Context Protocol (MCP) serves as a standard interface between large language models (LLMs) and external systems like PostgreSQL, emphasizing the importance of control over mere connectivity. An MCP server must act as a mediation and policy layer, enforcing security boundaries and ensuring that connections default to read-only. It should validate AI-generated SQL as untrusted input, encapsulate queries within transactions, and apply execution-time controls. The server must separate query generation from execution approval to manage operational costs and enforce guardrails on query types, such as limiting the number of rows returned. Token efficiency is crucial, with design considerations for compact data representation and schema introspection. In production, connection management is vital to prevent data leakage among multiple AI agents, and observability through logging executed queries and metadata is necessary for debugging and compliance. Long-running sessions require support for paginated responses to manage context effectively. Overall, the MCP server must integrate security, query safety, token efficiency, and observability into its design from the outset.