[{"content":"","date":null,"permalink":"https://kumarwrites.dev/topics/debian/","section":"topics","summary":"","title":"debian"},{"content":"I\u0026rsquo;ve always liked customizing my Linux desktop, but somehow, I\u0026rsquo;ve never paid much attention to the terminal. I\u0026rsquo;ve mostly stuck with the default Bash shell, which, while perfectly functional, isn\u0026rsquo;t particularly exciting to look at.\nRecently, I came across Oh My Zsh, a popular open-source framework for customizing the Zsh shell. It comes with a bunch of themes, plugins, and other features that let you personalize your command-line experience.\nI decided to give it a try on my Debian machine, and setting it up turned out to be pretty straightforward.\nWhat is Oh My Zsh? # Before getting started, there\u0026rsquo;s a small distinction worth understanding. Zsh (Z Shell) is a Unix shell, much like Bash, that offers several improvements, including better autocomplete, customization options, and interactive features.\nOh My Zsh, on the other hand, is a community-driven framework built on top of Zsh. It makes managing themes, plugins, and other shell configurations considerably easier.\nThink of it as a way to get more out of your existing terminal without having to configure everything from scratch. First, let\u0026rsquo;s get the fonts right # Before installing Oh My Zsh, there\u0026rsquo;s one more thing worth taking care of: your terminal font.\nSince I wanted to use the Agnoster theme, I needed a font that could display its special symbols and Powerline characters correctly. For this, I went with MesloLG Nerd Font.\nYou can download it from the Nerd Fonts website. Once downloaded, extract the archive and install the font files on your Debian machine.\nNext, open your terminal preferences and change the custom font to MesloLGM Nerd Font. The exact steps will depend on which terminal emulator you\u0026rsquo;re using. That\u0026rsquo;s it! Your terminal is now ready for any theme you throw at it.\nInstalling Oh My Zsh on Debian # First, we need to install Zsh, since Oh My Zsh depends on it. Thankfully, Debian has it available in its default repositories.\nOpen your terminal and run:\nsudo apt install zsh curl git Once installed, make Zsh your default shell:\nchsh -s \u0026#34;$(command -v zsh)\u0026#34; You\u0026rsquo;ll need to log out and back in for the change to take effect.\nNow comes the fun part. Installing Oh My Zsh is as simple as running this command:\nsh -c \u0026#34;$(curl -fsSL https://install.ohmyz.sh)\u0026#34; That\u0026rsquo;s it! The installer takes care of setting everything up, including creating the necessary configuration files in your home directory. Customizing the terminal with themes and plugins # This is where Oh My Zsh starts getting interesting. It comes with a collection of themes and plugins that you can enable through its configuration file, ~/.zshrc.\nThemes # Themes change how your shell prompt looks. You can find several built-in options, ranging from minimal prompts to more elaborate ones displaying Git information, directories, and other details.\nTo change your theme, open the configuration file:\nnano ~/.zshrc Look for the ZSH_THEME setting and replace its existing value with your preferred theme.\nFor example:\nZSH_THEME=\u0026#34;robbyrussell\u0026#34; Save the file and reload your configuration:\nsource ~/.zshrc Plugins # Plugins are another reason Oh My Zsh is worth exploring. They add useful functionality and shortcuts to your shell.\nYou can enable built-in plugins by adding their names to the plugins list in your .zshrc file.\nFor example:\nplugins=(git debian) Oh My Zsh also supports third-party plugins. Two particularly useful ones are zsh-autosuggestions and zsh-syntax-highlighting.\nThe former suggests commands based on your shell history as you type, while the latter adds syntax highlighting to your commands, making it easier to spot mistakes before executing them.\nYou can install both using Git:\ngit clone https://github.com/zsh-users/zsh-autosuggestions ${ZSH_CUSTOM:-~/.oh-my-zsh/custom}/plugins/zsh-autosuggestions git clone https://github.com/zsh-users/zsh-syntax-highlighting.git ${ZSH_CUSTOM:-~/.oh-my-zsh/custom}/plugins/zsh-syntax-highlighting Then add them to your plugins list:\nplugins=( git debian zsh-autosuggestions zsh-syntax-highlighting ) Make sure zsh-syntax-highlighting stays at the end of the list. Reload your configuration, and you\u0026rsquo;re good to go.\nIs Oh My Zsh worth trying? # I think so. If you spend a reasonable amount of time working in the terminal, having a shell that looks the way you want and offers a few additional conveniences is a nice improvement.\nOf course, Oh My Zsh isn\u0026rsquo;t going to completely change how you use Linux, nor do you need dozens of plugins to make it useful. Even a simple theme and a couple of plugins can make the experience considerably more pleasant.\nI\u0026rsquo;m still exploring its themes and plugins, but it\u0026rsquo;s been a nice little upgrade to my otherwise ordinary Debian terminal.\n","date":"20 November 2026","permalink":"https://kumarwrites.dev/writings/terminal-makeover-with-ohmyzsh/","section":"Writings","summary":"\u003cp\u003eI\u0026rsquo;ve always liked customizing my Linux desktop, but somehow, I\u0026rsquo;ve never paid much attention to the terminal. I\u0026rsquo;ve mostly stuck with the default \u003ca href=\"https://www.gnu.org/software/bash/\" target=\"_blank\" rel=\"noreferrer\"\u003eBash shell\u003c/a\u003e, which, while perfectly functional, isn\u0026rsquo;t particularly exciting to look at.\u003c/p\u003e","title":"I tried Oh My Zsh to give my Debian terminal a much-needed makeover"},{"content":"","date":null,"permalink":"https://kumarwrites.dev/","section":"Kumar's Blog","summary":"","title":"Kumar's Blog"},{"content":"","date":null,"permalink":"https://kumarwrites.dev/topics/ohmyzsh/","section":"topics","summary":"","title":"ohmyzsh"},{"content":"","date":null,"permalink":"https://kumarwrites.dev/topics/shell/","section":"topics","summary":"","title":"shell"},{"content":"","date":null,"permalink":"https://kumarwrites.dev/topics/terminal/","section":"topics","summary":"","title":"terminal"},{"content":"","date":null,"permalink":"https://kumarwrites.dev/topics/zsh/","section":"topics","summary":"","title":"zsh"},{"content":"","date":null,"permalink":"https://kumarwrites.dev/topics/database/","section":"topics","summary":"","title":"database"},{"content":"","date":null,"permalink":"https://kumarwrites.dev/topics/docker/","section":"topics","summary":"","title":"docker"},{"content":"I was looking for a simple way to organize my projects and ideas and maybe automate them with AI. Well the last part came up fairly recently during my experimentation but I needed something for that.\nI have used Airtable in the past and it was my go-to tool for collecting data and collaborating with others. However, I soon started to hit the limit of free tier and some important enough features that I needed were paywalled. That got annoying pretty quickly.\nSo I started searching for tools that were similar to Airtable but open-source, which I can easily self-host on my own hardware. And that’s how I stumbled upon NocoDB, and it turned out to be much more than I expected.\nBefore I dive too much into its features and capabilities, let me give you a brief overview of what it is.\nWhat exactly is NocoDB? # It’s an source-available, fair-code no-code platform that transforms your database into something that feels much like a familiar spreadsheet interface.\nInstead of having to write SQL queries or manually interact with the database, it gives a visual interface where you can create tables, add records, organize information and establish relationships between different datasets.\nUnder the hood, it connects directly to standard databases like SQLite, PostgreSQL, MySQL and SQL Server. If you already have an existing database running on your server, you can point NocoDB right at it.\nIs it really free? # The Community edition is free forever when self-deployed, with unlimited records, unlimited storage and unlimited seats. All five roles (owner, creator, editor, commenter, viewer), six view types, record comments and conditional webhooks are in the free tier.\nPaid plan exists for teams that need additional functionality like SSO, table and field-level permissions, audit logs and row-level security, which start at $24 per seat / month on a Business plan.\nFor a homelab or a small business, I believe the community edition gives plenty of features to get you started.\nSetup and installation with docker # Note: This section assumes that you already have Docker installed on your system and have some prior knowledge of compose files and mounted volumes . The fastest way to setup NocoDB is with SQLite database and a simple docker command:\ndocker run -d \\ --name noco \\ -v \u0026#34;$(pwd)\u0026#34;/nocodb:/usr/app/data/ \\ -p 8080:8080 \\ nocodb/nocodb:latest That’s it! Docker will create a container named noco, mount the nocodb bind mount for persistent data storage and expose the application on port 8080.\nThis setup is perfectly fine for getting started or running a lightweight installation. However, if you intend to use NocoDB regularly, especially with multiple users and larger datasets, I’d recommend using PostgreSQL as the underlying database.\nI prefer to write my own stack of services using docker compose files. Here’s a minimal two-container version with NocoDB and Postgres.\nFirst, create a separate directory that will hold our container stack and the bind mounts:\nAfter that, create the compose.yaml file:\nservices: nocodb: image: nocodb/nocodb:latest container_name: nocodb restart: unless-stopped ports: - \u0026#34;8080:8080\u0026#34; environment: - NC_DB=pg://nocodb-db:5432?u=noco\u0026amp;p=change-me\u0026amp;d=noco - NC_AUTH_JWT_SECRET=change-me-too volumes: - ./nocodb-data:/usr/app/data depends_on: - nocodb-db nocodb-db: image: postgres:16-alpine restart: unless-stopped environment: - POSTGRES_USER=noco - POSTGRES_PASSWORD=change-me - POSTGRES_DB=noco volumes: - ./pgdata:/var/lib/postgresql/data Before starting the containers, replace the example passwords and JWT secret with your own random values. Make sure the PostgreSQL password in NC_DB matches POSTGRES_PASSWORD.\nNow run:\ndocker compose up -d Give it a moment to pull the images and initialise the services. You can check whether everything is running with docker compose ps.\nOpen http://localhost:8080 in your browser, sign up, and you’re good to go. The first account you create automatically becomes super-admin.\nSome things you should know about:\nIf you skip the NC_DB variable, NocoDB quietly creates a SQLite database inside the volume instead, which is fine for starting-up. Set NC_AUTH_JWT_SECRET to something random. Put your service behind a reverse proxy with TLS before you open the port to the internet, because anyone who can read port 8080 gets a login page and can brute force themselves in. Feature that made me stay # Now that we’ve got NocoDB up and running, let’s talk about its features:\nSpreadsheets for everything # The first thing you will notice when opening NocoDB is its spreadsheet-like interface. You can create tables, add columns, enter records, rearrange fields, apply filters, sort information and group records according to different criteria.\nBut where it surpasses the spreadsheet is the variety of field types it offers. Apart from the usual text and number fields, you can create:\nDate and time fields Single-select and multi-select fields Checkboxes fields Attachment fields Formula fields (just like excel) Link fields (this one’s for creating relationships between tables) Lookup and rollup fields Different views make organising information much easier # Everyone has a preference when it comes to organizing data depending on the task. NocoDB tackle this issue by providing different Views for the same underlying table without duplicating data:\nGrid view: This is the traditional spreadsheet-style view. It’s particularly useful when you are working with large datasets, editing multiple records, or simply need a quick overview of everything in a table. Kanban view: Kanban organizes records into cards grouped by a particular field. For example you can group tasks according to their status and move them between columns as their progress changes. Gallery view: It displays records as cards, making it particularly useful when working with images, attachments, or visually oriented datasets. Calendar view: This one displays the records according to their date fields. Very useful for viewing which tasks are approaching their deadlines. Each view saves its own filters, sorts, grouping, and visible fields.\nNative forms for quick data capture # This one is particularly handy for collecting information, same as Google forms. NocoDB lets you generate standalone web forms directly from any table. You can pick which columns appear on the form, customize the labels, and share the link. Whenever someone submits the form, it drops straight into the database as a new row.\nYou can establish relationships between different tables # The most frustrating part of spreadsheets is repeating the same information across multiple tabs. NocoDB handles relational data effortlessly. You can link a “Projects” table to a “Tasks” using a simple “Link to Another Record” field.\nGranular permissions and collaboration # Even if you start out using NocoDB solo, you can easily invite teammates, editors, or clients later. It supports role-based access control, allowing you to lock down entire bases or restrict specific users to view-only or comment-only access.\nYou can also create password-protected public share links for specific views without giving outsiders access to the rest of the workspace.\nBuilt-in REST APIs and automation hooks # Every table created in NocoDB automatically generates its own REST API endpoints, complete with Swagger documentation built right into the dashboard.\nBetter yet, it plays nicely with automation tools like n8n, Activepieces, and Zapier. You can trigger webhooks whenever a row is created, updated, or deleted.\nAn MCP server, because it’s 2026 # NocoDB ships an MCP server; what this means is that you can connect compatible AI-clients interact with the data through natural-language instructions.\n\u0026gt; NOTE: Community Edition provides MCP access for reading and modifying records, while broader capabilities such as managing tables, fields, and workflows requires a licensed deployment.\nMy thoughts # Airtable still has the better interface. I’m not gonna pretend otherwise, and if you need a base for two people and a few hundred rows, the free tier is fine. Stay where you are.\nFor me, NocoDB is essentially just a fancy wrapper around my data. The reason I chose it, if this project were to disappear tomorrow, or if I decided to migrate to a completely different frontend three years from now, I would not be stuck exporting messy CSV files or annoying proprietary backups.\nWould I recommend it? If you already run Docker for something else, yes, without much hesitation. If you have never touched a server before then No because you become the sysadmin yourself and that means you have to look after the updates, periodic backups, TLS certificates and the restore process.\nIf you have already subscribed to some other platform, do what I did. Spin it up on a spare evening, move one table over, and test it for a week before you cancel anything.\n","date":"20 November 2026","permalink":"https://kumarwrites.dev/writings/nocodb-self-hosted-airtable-alternative/","section":"Writings","summary":"\u003cp\u003eI was looking for a simple way to organize my projects and ideas and maybe automate them with AI. Well the last part came up fairly recently during my experimentation but I needed something for that.\u003c/p\u003e","title":"I turned my database into a spreadsheet with NocoDB and it's surprisingly good"},{"content":"","date":null,"permalink":"https://kumarwrites.dev/topics/mysql/","section":"topics","summary":"","title":"mysql"},{"content":"","date":null,"permalink":"https://kumarwrites.dev/topics/postgres/","section":"topics","summary":"","title":"postgres"},{"content":"","date":null,"permalink":"https://kumarwrites.dev/topics/sqlite/","section":"topics","summary":"","title":"sqlite"},{"content":"Linux Mint has stirred things up in a major way. After years of following a predictable six-month release cycle with alphabetical girls\u0026rsquo; names, the project has officially shifted gears.\nMoving forward, users will see one major release every year with no point releases in between, named sequentially using boys\u0026rsquo; names.\nBuilt on top of the Ubuntu 26.04 LTS base and backed by a five-year support period. Linux Mint 27 \u0026ldquo;Adrien\u0026rdquo; slated for release around Christmas Eve, this update packs some of the most ambitious changes the desktop environment has seen in years.\nHere is a detailed breakdown of what is landing in your Christmas stocking.\nNote: Linux Mint 27 is still under development. The features discussed below are based on the development team\u0026rsquo;s announcements and may change before the stable release. 1. Wayland Is Finally Ready for Prime Time # Both X11 and Wayland will be fully supported in the upcoming Cinnamon release, with Wayland performance now sitting nearly level with X11.\nWindows and menus map properly, focus stealing is prevented, and crashes across Cinnamon, Muffin, cinnamon-session, and Xwayland have been addressed.\nMulti-monitor setups and KVM switches are fully supported, and hardware acceleration gets a boost with GBM over EGL for NVIDIA cards alongside HiDPI scaling for cursors and Chromium-based apps like Slack and VS Code.\nThe final piece of the puzzle was the screensaver. The new screensaver is rendered directly by the compositor and managed by Cinnamon rather than a separate process, ensuring seamless operation across both Wayland and X11.\n2. Desktop Upgrades and File Management # Nemo File Manager Gets a Preview Pane # The default file manager, Nemo, gains a built-in preview pane that can be toggled using the F7 key or directly from the status bar.\nIt handles media files like audio, pictures, and video while displaying detailed metadata such as camera models and capture dates.\nInteractive search also gets an upgrade, turning into a real-time filtered view that narrows down results as you type rather than just jumping to the first match.\nDedicated Screenshot Utility # Cinnamon finally gets a dedicated screenshot tool. It supports capturing single windows (with or without shadows), multi-monitor layouts, or specific regions. Cropping tools are built right in before saving or copying to the clipboard.\nLibrary Redesign (Thingy) # The desktop document manager, Thingy, gets a redesign that groups recent files by document type instead of the application used to open them.\nA customizable sidebar lets you reorder or hide groups, and you can define custom categories based on file types and directories.\n3. System and Installer Changes # A Unified Live Installer # Linux Mint is moving away from Ubiquity, adopting the modern live installer already used in Linux Mint Debian Edition (LMDE). The new installer streamlines OEM installations, BIOS/EFI handling, Secure Boot, and advanced LVM or LUKS disk setups.\nThe Return of Hibernation # After being dropped upstream by Ubuntu, hibernation makes a welcome return via a new \u0026ldquo;Sleep Modes\u0026rdquo; section in the System Administration tool.\nThe panel also introduces Hybrid Sleep (which suspends to RAM while writing the session to disk) and Suspend-then-Hibernate (which switches to hibernation if battery levels hit critical thresholds).\nCentralized Kernel Management # Kernel management has been pulled out of the Update Manager and integrated into the System Administration tool for both Mint and LMDE.\nYou can track specific kernel series, configure retention rules for older kernels, and let automated weekly cleanups handle the maintenance.\n4. Brand New Core Applications # Linux Mint 27 introduces several new in-house XApps designed to reduce dependency on GNOME infrastructure:\nSuperlight: A fresh system monitor tracking CPU load, memory, swap, GPU usage, network traffic, disk I/O, fan speeds, and temperatures, alongside a dedicated battery status page. Velocitty: A modern default terminal emulator featuring tab overviews inspired by Ghostty and Ptyxis, customizable color palettes, and process-aware tab color coding. Xepub: A dedicated EPUB ebook reader with chapter navigation, bookmarks, zoom controls, and configurable online dictionary lookup engines. Calendar \u0026amp; Clocks: A new suite from the Clockenstein project. The Calendar app syncs with local databases, Google Calendar, and CalDAV servers like Nextcloud. Clocks brings an Android-inspired alarm and timer system with custom recurring sounds. Unfold: A lightweight archive manager backed by 7zip and 7zip-RAR for handling compressed and password-encrypted archives. 5. Miscellaneous Improvements # Before release day arrives, a few smaller enhancements are rounding out the feature set:\nNative WPA3 and OWE Wi-Fi security support in Cinnamon. Fingerprint login support in Slick Greeter without requiring an extra Enter keystroke. An automated \u0026ldquo;Sensors\u0026rdquo; page in System Reports. A comprehensive Mint-Y theme refresh featuring improved dark mode contrast, rounded window frames, and refined list views. When Can You Download It? # Linux Mint 27 \u0026ldquo;Adrien\u0026rdquo; is targeted for release around December 24, 2026, and will be available in Cinnamon, MATE, and Xfce editions. Until then, adventurous users can pull and build components directly from source.\n","date":"12 November 2026","permalink":"https://kumarwrites.dev/news/linux-mint-27-adrien-features/","section":"News","summary":"\u003cp\u003e\u003ca href=\"https://www.linuxmint.com/\" target=\"_blank\" rel=\"noreferrer\"\u003eLinux Mint\u003c/a\u003e has stirred things up in a major way. After years of following a predictable six-month release cycle with alphabetical \u003cem\u003egirls\u0026rsquo;\u003c/em\u003e names, the project has officially shifted gears.\u003c/p\u003e","title":"Linux Mint 27 'Adrien' Arrives This Christmas With Full Wayland Support and New Core Apps"},{"content":"","date":null,"permalink":"https://kumarwrites.dev/topics/news/","section":"topics","summary":"","title":"news"},{"content":"The Linux kernel development cycle has been getting increasingly busy, but Linux 7.4 is taking things to another level.\nWith its merge window now closed and the first release candidate already out, this upcoming kernel version has managed to set a new record for the number of commits merged during a single merge window.\nAccording to Phoronix, Linux 7.4 received 19,671 commits during its merge window, beating the previous record of 16,267 commits set by Linux 7.3. That\u0026rsquo;s quite a jump, especially considering that a typical merge window usually sees somewhere around 13,000 - 15,000 commits.\nPersonally, I find this release interesting for reasons beyond its impressive commit count.\nSo, here are some of the Linux 7.4 features and changes that I find worth mentioning.\nFeatures # AMD Ryzen Zen 6 Gets More Linux Support - More hardware enablement, power management improvements, and support for the new BTB CTX security feature.\nAMD Radeon HDMI 2.1 Enabled by Default - AMDGPU now enables HDMI 2.1 FRL by default, alongside VRR, FreeSync, and ALLM support.\nAMD PerfOpt Promises Better AI Performance on Radeon iGPUs - A new IOMMU optimization can improve AI/LLM performance on supported AMD integrated GPUs, with early benchmarks showing gains of up to 18- 23% in certain workloads.\nInitial Apple M4 and A18 Pro Support Arrives - Mainline Linux can now boot on supported Apple M4 and A18 Pro devices, although hardware functionality remains very limited.\nValve\u0026rsquo;s Steam Frame VR Headset Gets Mainline Support - Initial Device Tree support for Valve\u0026rsquo;s upcoming VR headset has been merged.\nNintendo Switch 2 Controllers Get a New Linux Driver - A new HID driver adds initial input support for Joy-Con 2, Pro Controller 2, and GameCube NSO controllers, thanks to work sponsored by Valve.\nGoogle Tensor G5 and Pixel 10 Support Goes Mainline - Linux 7.4 brings initial upstream support for Google\u0026rsquo;s latest Tensor SoC and Pixel 10 devices.\nIntel Xe Gets VRAM Health Checks - Intel\u0026rsquo;s graphics driver gains video memory health checks and improved handling of degraded VRAM. Qualcomm\u0026rsquo;s MSM DRM driver also adds support for newer Adreno GPUs.\nLinux 7.4 Prepares for CXL Dynamic Capacity Devices - Initial groundwork for dynamically expanding and reducing server memory capacity without rebooting the host.\nMajor Performance Improvements to Memory Management and Workqueues - A memblock optimization reduces a specific memory retrieval test from hundreds of seconds to milliseconds, while workqueue changes bring up to 10 - 100x faster idle flushes on large systems. (More details).\nShould You Upgrade? # Welp, that\u0026rsquo;s a personal question and I\u0026rsquo;ll leave that for you to decide. :P\nOn a more serious note, Linux 7.4 is shaping up to be an interesting release. Linus Torvalds has also acknowledged the unusually large merge window. Interestingly, he believes AI-assisted development is one of the major reasons behind the increase.\nMore AI-assisted bug reports, fixes, and suggested patches are making their way into kernel development. He also mentioned that the possibility of Linux 7.4 becoming the next LTS release might have contributed to the additional development activity.\nWhether Linux 7.4 eventually becomes the next LTS release remains to be seen.\nAs of November 4, Linux 7.4 is in the release-candidate stage, with v7.4-rc1 available. So, there is still some testing and stabilization work ahead before the final release.\nPersonally, I\u0026rsquo;d suggest waiting for the stable release unless you have a specific reason to test it early.\n","date":"4 November 2026","permalink":"https://kumarwrites.dev/news/linux-7-4-kernel-features/","section":"News","summary":"\u003cp\u003eThe Linux kernel development cycle has been getting increasingly busy, but Linux 7.4 is taking things to another level.\u003c/p\u003e","title":"Linux 7.4 Brings Biggest Merge Window, AMD Zen 6 Support \u0026 Next LTS"},{"content":"A new Linux kernel update has officially landed. Arriving right on schedule about two months after Linux 7.2, version 7.3 brings a solid mix of performance speedups, low-VRAM gaming fixes, and groundwork for future hardware.\nAs usual, Linus Torvalds shared a quick update with the community before heading out on his annual family vacation.\nHe mentioned that he will be spending time on a boat and scuba diving underwater while the merge window for Linux 7.4 opens up.\nLet us take a closer look at what is new in Linux 7.3.\nStorage and Filesystem Upgrades # For those of us running self-hosted servers or homelabs, the filesystem updates are incredibly exciting.\nBtrfs gets a major speed boost: Early testing shows some massive performance improvements for Btrfs along with filesystem cleanups and changes to free-space handling compared to 7.2. Database performance: There is new RWF_DONTCACHE support for block devices, which is a huge win if you run database servers directly off raw devices. Faster NVMe: A known bottleneck for small I/O operations on PCIe Gen5 NVMe SSDs has finally been addressed. Apple Time Machine: The in-kernel SMB server, KSMBD, now fully supports interoperability for Apple Time Machine backups. FUSE improvements: FUSE gets a nice performance bump thanks to IO_uring buffer pools and zero-copy handling. FailFS is here: The new FailFS driver was officially merged, complementing the NULLFS driver from the last cycle. Security fixes: eCryptfs received some much needed security patches, with a few fixing issues that date all the way back to its introduction in 2006! Graphics and Gaming Performance # Valve\u0026rsquo;s Linux graphics team has been putting in the work, and gamers are reaping the benefits.\nLow VRAM optimizations: If your GPU only has around 8GB of video memory, the kernel can now intelligently move less important data out to your system RAM. In testing, this allowed demanding games like Indiana Jones and the Great Circle to run smoothly on 8GB cards even when asking for 9GB. AMD GPU crash recovery: Older Radeon cards get much better recovery from GPU hangs. The kernel now only resets the GFX block and leaves your graphics memory intact. Intel Xe3P Nova Lake: Integrated graphics for these chips are now considered stable and are enabled by default. Nouveau updates: The open source Nvidia driver gains NVDEC interface support, which is needed for Vulkan Video acceleration with Mesa 26.3. Retro hardware: The old framebuffer device drivers (FBDEV) got improvements for vintage Atari computers and 3dfx Voodoo 3, 4, and 5 graphics cards. Processors and Memory Management # There are some fantastic memory optimizations in this release that will directly benefit low-cost SBCs and older hardware.\nCrucial memory lock fixes: ZTE engineers fixed a KSM lookup bug that could freeze apps for hundreds of milliseconds under memory pressure, bringing the delay down from 705ms to under 2ms. Xiaomi engineers also tackled a zsmalloc lock slowdown, yielding up to a 1.83x performance gain on the Raspberry Pi 4B. AMD Zen 6 preparations: AMD has been incredibly busy laying the groundwork for next-gen Zen 6 platforms, including adding protocol v7 support to the HSMP driver. Apple Silicon: The Apple M3 Pro, Max, and Ultra chips now have initial support for booting Linux, while the older M1 Pro, Max, and Ultra SoCs received some new power optimizations. New Architectures: We are seeing initial enablement for IBM Power12 (which also begins supporting in-kernel Rust code), Qualcomm Shikra SoC support, and initial prep for Intel Starfire, a space-grade chip built for extreme environments. Dynamic EPP: If you like tweaking CPU power, you can now toggle the Dynamic Energy Performance Preference per CPU core instead of just system wide. Peripherals and Extras # Rounding out the release are a few neat hardware additions:\n2026 Steam Controller: Initial kernel driver support has been added for the upcoming 2026 Steam Controller. Ryzen AI Halo LEDs: Mini PC owners can now natively control the RGB lighting on their Ryzen AI Halo devices directly through the kernel. ThinkPad Security: Users with Lenovo ThinkPads can finally take advantage of the USB-C Restricted Mode Security functionality on Linux. Logitech peripherals: You can now remap buttons on newer enthusiast class Logitech mice and keyboards using HID++ 2.0. Getting Linux 7.3 # This is easily one of the most feature packed releases of the year. If you want to try it out immediately, the source code is available right now on kernel.org. Otherwise, you can expect it to land very soon in rolling release distributions like Arch Linux and Fedora Workstation, or by installing the latest Ubuntu 26.10 interim release.\nWhat are your thoughts on all the AI generated code making its way into the kernel? Let me know down in the comments!\n","date":"20 October 2026","permalink":"https://kumarwrites.dev/news/linux-kernel-7-3-release/","section":"News","summary":"\u003cp\u003eA new Linux kernel update has officially landed. Arriving right on schedule about two months after Linux 7.2, version 7.3 brings a solid mix of performance speedups, low-VRAM gaming fixes, and groundwork for future hardware.\u003c/p\u003e","title":"Linux Kernel 7.3 Released with Faster I/O, Better Gaming \u0026 More"},{"content":"Hey everyone, it is that time of the year again. Canonical just dropped Ubuntu 26.10, officially codenamed \u0026ldquo;Stonking Stingray\u0026rdquo;, and it is shaping up to be a pretty wild interim release.\nSince this is an October release, it only gets nine months of support (until July 2027). But if you like living on the edge and want newer packages than what the 26.04 LTS offers, there is a lot to dig into here.\nI spent some time checking out the official release notes, and here is a breakdown of what makes this update actually worth your time.\nThe Rustification is Complete # For a while now, Ubuntu has been slowly oxidizing its core tools. With 26.10, the migration to the Rust-based uutils implementation is totally finished.\nPreviously, commands like cp, mv, and rm stayed on the GNU versions because of compatibility hiccups. Now, those are officially running on Rust.\nThe memory safety push doesn\u0026rsquo;t stop there. Canonical also brought Sequoia PGP into the main archive to replace GnuPG for verifying APT repository signatures. Built by former GnuPG devs, Sequoia brings its sq and sqv tools right alongside the classic gpg, acting as a barrier against future quantum cryptographic attacks.\nLinux 7.3 and Big Gaming Wins # Under the hood, Stonking Stingray is powered by the shiny new Linux 7.3 kernel. If you are rocking one of those newer ARM laptops, you will be happy to know this kernel brings solid support for Snapdragon X2 Elite chips, plus older X Elite and X Plus models from Lenovo, HP, and Acer.\nFor the gamers out there, this release includes the first round of Valve engineer Natalie Vock’s VRAM overcommit work. Basically, if your graphics card is low on video memory, some of that data can now be offloaded to your regular system RAM.\nCombine that with the new Mesa 26.2 graphics drivers and the shift to the NVIDIA 615 driver series (which ships open kernel modules only for Turing and newer GPUs), and you have a heavily performance-focused setup.\nGNOME 51 and Desktop Tweaks # Ubuntu 26.10 ships with GNOME 51, which brings a ton of quality of life improvements. You get a new fingerprint manager in the Users panel, DNS search domains right inside the Network settings, and a neat little count badge in Files when you drag multiple items around. The built-in Ubuntu Tiling Assistant also got an update, offering a super clean interface for window snapping.\nI also noticed a really nice technical change while doing some extra reading. Ubuntu 26.10 adopted dbus-broker as the default D-Bus implementation. This replaces the old legacy message queuing with strict per-app accounting, meaning one misbehaving app can no longer stall your entire desktop.\nMeet Myna: The Local AI Agent # Here is something the initial leak roundups missed. Canonical is testing out a brand new, opt-in speech-to-text agent called Myna. Named after the bird that mimics human speech, Myna is designed to process everything locally so your privacy stays intact.\nRight now, it is an early look at intent-driven computing. Instead of clicking through five menus, a developer could just say they want to troubleshoot a failing test, and the OS would prepare a workspace. It is definitely experimental at the moment, but Canonical plans to have it fully baked by the 28.04 LTS release.\nHomelab and Security Upgrades # Being a homelab guy, I always look at the security and server-side bumps.\nOpenSSL 4 is here, bringing Encrypted Client Hello (ECH) and better QUIC support. TPM-backed full-disk encryption now works even if your hardware lacks a root of trust, as long as you set a PIN or passphrase. For the RISC-V fans, there is official RVA23 support for SpacemiT K3 boards and SiFive\u0026rsquo;s BigSky platform, plus brand new RISC-V desktop images. The developer toolchain got bumped to GCC 15.2.0, LLVM 22.1.8, Python 3.14.7, Go 1.27, and OpenJDK 25.0.4. Wrapping Up # If you are ready to tinker, you can grab the AMD64, ARM64, or RISC-V ISOs from the official Ubuntu download page. Let me know in the comments if you plan on daily driving this or if you are safely sticking to the LTS branch!\n","date":"18 October 2026","permalink":"https://kumarwrites.dev/news/ubuntu-stonking-stingray-out/","section":"News","summary":"\u003cp\u003eHey everyone, it is that time of the year again. Canonical just dropped Ubuntu 26.10, officially codenamed \u0026ldquo;Stonking Stingray\u0026rdquo;, and it is shaping up to be a pretty wild interim release.\u003c/p\u003e","title":"Ubuntu 26.10 'Stonking Stingray' is Out and Its Packed with Rust"},{"content":"In a major shift for one of the most popular Linux distributions, Linux Mint is completely overhauling how it builds, names, and releases its operating system.\nIf you were waiting for Linux Mint 23, you can stop holding your breath, the next major release will be Linux Mint 27, codenamed \u0026ldquo;Adrien\u0026rdquo;.\nAnnounced by project lead Clement Lefebvre in the September 2026 monthly update, the changes mark the end of a long decision-making process regarding the distribution\u0026rsquo;s future. Here is a breakdown of exactly what is changing and why.\nThe Big Changes Coming This Christmas # Targeting a Christmas 2026 release, Linux Mint 27 \u0026ldquo;Adrien\u0026rdquo; will be built on the Ubuntu 26.04 LTS base. But the underlying architecture isn\u0026rsquo;t the only thing getting refreshed. Mint is throwing out its old playbook:\nA Jump to Version 27: Developers temporarily used \u0026ldquo;23\u0026rdquo; and \u0026ldquo;Alfa\u0026rdquo; earlier this year while ironing out their strategy. They ultimately decided the version number will now align with the upcoming calendar year. Because the new OS drops around Christmas, the vast majority of its lifecycle will take place in 2027, making \u0026ldquo;27\u0026rdquo; the logical choice. The End of Point Releases: Say goodbye to fractional version numbers like 22.1 or 22.2. Mint is moving to a strict annual release cycle, meaning version 27 will be followed directly by version 28 next year. This simplified upgrade path is designed to make things easier for beginners transitioning to Linux. A New Naming Convention: Since 2006, Mint has used feminine codenames, cycling alphabetically from version 1.0\u0026rsquo;s \u0026ldquo;Ada\u0026rdquo; all the way to 22.3\u0026rsquo;s \u0026ldquo;Zena\u0026rdquo;. Having literally run out of the alphabet just as the release schedule changed, the team is starting fresh with masculine names, kicking things off with \u0026ldquo;Adrien\u0026rdquo;. Why the Shake-Up? # It all comes down to developer bandwidth. Previously, the team was juggling two major releases a year alongside maintaining Linux Mint Debian Edition (LMDE). This frantic six-month cycle left them with little room to actually step back and build substantial new features.\nBy slowing down to a 12-month cycle, the Mint team finally has the breathing room to tackle ambitious new projects that would have previously felt out of reach.\nWhat to Expect in \u0026ldquo;Adrien\u0026rdquo; # We\u0026rsquo;re already seeing the fruits of this longer development cycle. The Clockenstein project, which gave Mint a new calendar agent earlier this year, is expanding with a brand new Clocks app.\nBorrowing design cues and behaviors from Android, it handles recurring alarms, custom sounds, and pushes alerts through a background daemon even when the application is closed.\nThe team is also putting the finishing touches on three more mystery projects Superlight, Unfold, and Velocitty which are expected to be unveiled in the coming months.\nLinux Mint 27 promises to be a massive reset for the project. By cutting out the halfway-point updates and giving developers a full year to polish the desktop, \u0026ldquo;Adrien\u0026rdquo; might just be the most feature-rich Mint release we\u0026rsquo;ve seen in years.\n","date":"14 October 2026","permalink":"https://kumarwrites.dev/news/linux-mint-23-dead/","section":"News","summary":"\u003cp\u003eIn a major shift for one of the most popular Linux distributions, \u003ca href=\"https://www.linuxmint.com/?utm_source=kumarwrites.dev\" target=\"_blank\" rel=\"noreferrer\"\u003eLinux Mint\u003c/a\u003e is completely overhauling how it builds, names, and releases its operating system.\u003c/p\u003e","title":"Linux Mint 23 is Dead: Enter Mint 27 'Adrien' and a Brand New Release Cycle"},{"content":"In a striking reversal of his past skepticism, Linux creator Linus Torvalds has not only admitted he enjoys using AI, he\u0026rsquo;s now shipping kernel code built with its help.\nA new branch in Torvalds\u0026rsquo; own kernel repository, named noref-filemap-read-with-llm, contains an eight-patch series built on Linux 7.3-rc6 plus fixes.\nWhile the commit message offers no target release for merging, its contents mark a significant milestone. The commit message itself credits Claude Code, working under Torvalds\u0026rsquo; direction, for assisting with the code, test programs, a draft of the commit message, and even the review behind its list of known problems.\nWhat did he actually FIX? # The patch series addresses a performance bottleneck that dates back to 2024, when Matthew Wilcox flagged the issue based on a customer\u0026rsquo;s workload.\nWhen many threads read the same file, the kernel serves them from the page cache, its in-memory copy of recently read data. Currently, each read takes a hold of the cached data and releases it afterward. Every thread updates the same counter to do so, causing threads to trip over each other.\nTorvalds\u0026rsquo; patches skip that step entirely. To keep it safe, memory leaving the cache isn\u0026rsquo;t handed out for reuse until a short period has passed, allowing a reader to copy data directly without holding anything. The trade-off: more memory sits idle briefly.\nThe payoff, measured on Torvalds\u0026rsquo; own laptop with some security protections disabled, is dramatic—eight threads reading one file jumped from about 10 million reads per second to between 30 and 33 million.\n\u0026ldquo;I Actually Really Like Using AI\u0026rdquo; # The kernel work follows comments Torvalds made Friday at Open Source Summit Europe in Prague, where he spoke with longtime friend Dirk Hohndel, head of Ericsson Software Technology.\nZDNet\u0026rsquo;s Steven Vaughan-Nichols, who attended, quotes Torvalds saying, \u0026ldquo;I actually really like using AI.\u0026rdquo;\nTorvalds said he used to think AI wasn\u0026rsquo;t very good but has changed his mind. He now views it as a tool that works well when treated as such—and one that makes programming more enjoyable for him.\nHis personal example: a guitar pedal project with an interface he\u0026rsquo;d written in C. Wanting a more modern look, he turned to AI. \u0026ldquo;I use AI to do the things that I\u0026rsquo;m bad at,\u0026rdquo; he said.\nHe also defended vibe coding as \u0026ldquo;a wonderful way to find joy in programming,\u0026rdquo; particularly for newcomers. But he acknowledged a growing concern: while AI improves the code base, the load it places on maintainers has become a real problem.\nA Notable Shift # This isn\u0026rsquo;t Torvalds\u0026rsquo; first dance with AI on kernel work. In August, he used it to help fix a bug in the Intel Xe graphics driver.\nBut the public embrace represents a remarkable evolution. In 2024, Torvalds dismissed the AI industry as \u0026ldquo;90 percent marketing and 10 percent reality.\u0026rdquo; Now, he\u0026rsquo;s actively using it for kernel work—and crediting it in his commits.\nFor a developer known for his blunt assessments and hands-on approach, the message is clear: AI has earned a place in the kernel workflow, at least in Torvalds\u0026rsquo; branch.\n","date":"13 October 2026","permalink":"https://kumarwrites.dev/news/linus-torvalds-likes-claude/","section":"News","summary":"\u003cp\u003eIn a striking reversal of his past skepticism, Linux creator Linus Torvalds has not only admitted he enjoys using AI, he\u0026rsquo;s now shipping kernel code built with its help.\u003c/p\u003e","title":"Linus Torvalds Embraces AI, as His New Commit Message Credit Claude"},{"content":"Brief notes and updates on Linux releases, security alerts, and open-source developments that caught my attention.\n","date":null,"permalink":"https://kumarwrites.dev/news/","section":"News","summary":"\u003cp\u003eBrief notes and updates on Linux releases, security alerts, and open-source developments that caught my attention.\u003c/p\u003e","title":"News"},{"content":"","date":null,"permalink":"https://kumarwrites.dev/topics/ai/","section":"topics","summary":"","title":"ai"},{"content":"For a good while, I have been hearing a lot of hype about AI agents on my social media feed e.g. \u0026ldquo;AI agent that book your flight\u0026rdquo; or \u0026ldquo;agents that refactor your codebase\u0026rdquo;, or the most interesting one \u0026ldquo;agent that does your weekly shopping for you\u0026rdquo;.\nI was kinda skeptical it at first as it all sounded like fancy marketing until I instaslled Hermes Agent myself.\nSo What Even Is Hermes? # Hermes Agent is a self-hosted, open-source AI agent from Nous Research. The pitch is that it\u0026rsquo;s the only agent with a built-in learning loop.\nIt creates reusable skills from the tasks it completes, refines those skills as it uses them, keeps a curated memory of who you are and what you\u0026rsquo;re working on, and can search its own past conversations to recall things from weeks ago. What I\u0026rsquo;m Using? # Note: Hermes is model-agnostic, so you can easily swap models out for OpenAI, Anthropic, OpenRouter, Google, xAI, or run local inference through Ollama or LM Studio. For this tutorial, I am using a RackNerd VPS with docker installed on it and for LLM, I\u0026rsquo;m using the DeepSeek API. It\u0026rsquo;s cheap \u0026amp; smart enough for agentic work.\nHere is the thing about AI agents: they run shell commands, read files, and reach out to the internet. That is the product.\nBut most people install one directly on their machines, hand it a terminal, and have no real idea what it does under the hood. You are not auditing every tool call. Nobody is.\nIf you are just experimenting, that is fine, but do it sensibly: run the agent in a VM, or better, a Docker container and that\u0026rsquo;s what we gonna do in this tutorial.\nRequirements # Docker \u0026amp; Docker Compose Step 1: Prep the directory \u0026amp; pull the image # Create a new directory for the project and move into it:\nmkdir -p hermes \u0026amp;\u0026amp; cd hermes Before writing any configuration, pull the latest image to get the heavy lifting out of the way:\ndocker pull nousresearch/hermes-agent:latest Step 2: Write your compose file # I use docker compose because it is version-controllable and saves you from remembering twelve-flag CLI commands.\nOpen a new file with your editor of choice:\nnano docker-compose.yaml Paste in the configuration, making sure to swap in your own secure passwords and secrets:\nservices: hermes: image: nousresearch/hermes-agent:latest container_name: hermes restart: unless-stopped command: gateway run volumes: - ~/.hermes:/opt/data # Maps your local data into the container environment: - HERMES_DASHBOARD=1 # Mandatory Auth Variables - HERMES_DASHBOARD_BASIC_AUTH_USERNAME=admin - HERMES_DASHBOARD_BASIC_AUTH_PASSWORD=choose-a-strong-password - HERMES_DASHBOARD_BASIC_AUTH_SECRET=your-random-32-char-secret-string # LLM Provider Key - DEEPSEEK_API_KEY=your-deepseek-api-key-here ports: - \u0026#34;8642:8642\u0026#34; # gateway API - \u0026#34;9119:9119\u0026#34; # web dashboard I\u0026rsquo;m setting up my DeepSeek API key in the environment variables, but if you want, you can do that from the web-ui as well. Step 3: Boot the agent # Because our API key and authentication secrets are already defined in the compose file, you do not need to run any preliminary setup commands.\nJust fire it up in the background:\ndocker compose up -d If something feels off or you want to verify your configuration, you can easily check the logs with:\ndocker logs hermes OR you can run the agent\u0026rsquo;s built-in diagnostic tool directly inside the running container using Docker\u0026rsquo;s interactive mode:\ndocker exec -it hermes hermes doctor This checks for missing dependencies, misconfigured keys, and network issues without interrupting active sessions.\nStep 4: Access the Dashboard # Open your browser and head to http://\u0026lt;your-vps-ip\u0026gt;:9119.\nBecause we\u0026rsquo;ve already set the auth variables, we\u0026rsquo;ll will be welcomed by the Hermes Agent login page. Sign in using the exact USERNAME and PASSWORD you defined in your docker-compose.yaml. Once authenticated, you land in the Hermes Web UI.\nAt first glance, it might looks overwhelming but here is how the interface breaks down:\nChat: The main interface where you talk directly to the agent and watch its live reasoning loops and tool executions. Models: The hub for managing your LLM providers, swapping models, or pointing the agent to local endpoints. Files: A built-in file explorer to inspect, upload, or download whatever the agent is working on inside its sandbox. Cron: The scheduling engine where you set up recurring background tasks and daily automations. Skills: The library where Hermes stores reusable code and custom tools it has learned or that you have enabled. Plugins: The toggle board for enabling extra capabilities and third-party extensions. MCP: (Model Context Protocol) The menu for plugging in external tool servers, like GitHub, Home Assistant, or your local databases. Channels: Your gateway manager for setting up outside integrations like Telegram, Discord, Slack, and IMAP Email. Webhooks: Where you configure incoming webhooks so external services can trigger the agent. There are a few other settings and menus to dig into, but you can explore those further yourself, as I am honestly still learning about them too.\nA Word on Integrations # One of the biggest reasons to run Hermes is its messaging gateway. It can talk to you wherever you already are, the platform list is genuinely absurd. Here\u0026rsquo;s the capability matrix for the major ones:\nPlatform Voice Images Files Threads Typing Streaming Telegram ✅ ✅ ✅ ✅ ✅ ✅ Discord ✅ ✅ ✅ ✅ ✅ ✅ Slack ✅ ✅ ✅ ✅ ✅ ✅ WhatsApp — ✅ ✅ — ✅ ✅ Signal — ✅ ✅ — ✅ ✅ Email — ✅ ✅ ✅ — — Matrix ✅ ✅ ✅ ✅ ✅ ✅ Microsoft Teams — ✅ — ✅ ✅ — Plus SMS, Google Chat, Mattermost, DingTalk, Feishu/Lark, WeCom, Weixin, QQ, LINE, ntfy, IRC, SimpleX, Home Assistant, and more.\nSetting these up requires passing credentials and managing session pairings, which deserves its own focus. We will cover how to hook Hermes up to your inbox and chat apps in the next tutorial.\nConclusion # I believe this guide serves as a good starting point for anyone just beginning to venture into the realm of agentic AI. The reason I chose docker is because it saves me from the dependencies hell and it\u0026rsquo;s a great way to isolate your system\u0026rsquo;s attack surface area if the agent happens to break loose. XD\nHappy tinkering!\n","date":"12 October 2026","permalink":"https://kumarwrites.dev/writings/setup-hermes-with-docker/","section":"Writings","summary":"\u003cp\u003eFor a good while, I have been hearing a lot of hype about AI agents on my social media feed e.g. \u003cem\u003e\u0026ldquo;AI agent that book your flight\u0026rdquo;\u003c/em\u003e or \u003cem\u003e\u0026ldquo;agents that refactor your codebase\u0026rdquo;\u003c/em\u003e, or the most interesting one \u003cem\u003e\u0026ldquo;agent that does your weekly shopping for you\u0026rdquo;\u003c/em\u003e.\u003c/p\u003e","title":"How to Setup Hermes Agent with Docker"},{"content":"","date":null,"permalink":"https://kumarwrites.dev/topics/vps/","section":"topics","summary":"","title":"vps"},{"content":"Me # Hi there! I\u0026rsquo;m Abhishek, a tech writer and a journalist based in India. I am passionate about computers and technology.\nIn my free time I fiddle around with a lot of stuff. It’s my motto to fix everything that can be fixed/ saved or at the very least salvaged from especially in an era dominated by “planned obsolescence”.\nApart from all of this, I like to write with literal pen \u0026amp; paper. I don\u0026rsquo;t do daily journaling but more like twice a week or once in 3 months. XD\nOh, and I also like to collect fountain pens. They are so cool. I use them daily. The idea of simple things that can outlast their owners, just amuses me.\nPlease contact me if you want to work together!\nThe blog # This is my personal blog, as a strong advocate for digital ownership \u0026amp; privacy I primarily write about self-hosting solutions, open-source softwares, low-cost hardwares, SBCs, system administration and little bit about web-dev.\nI have worked in two major publications It’s FOSS \u0026amp; Linux Handbook, you can read my work there. That led to the start of this blog.\nI consider this blog a successor to my Kumar Projects blog, which I started in 2019. It taught me a ton about running websites, CMS, TLS certs \u0026amp; SEO.\nAlthough, this blog is:\nBuilt with Hugo, an open-source static site generator\nComments are powered by Isso\nLicensed under CC BY-NC 4.0.\nContact # The best way to reach me:\nComments on blog posts for public discussions\nGitHub issues for code-related matters\nEmail: \u0026lt;hello@kumarwrites.dev\u0026gt; or social media for everything else\nPlease use public channels like comments or GitHub issues when possible - it helps others learn too!\nSupport # If you\u0026rsquo;d like to support my work and public services:\nbuymeacoffee.com/abhishk\nThanks for visiting.\n","date":"3 October 2026","permalink":"https://kumarwrites.dev/about/","section":"Kumar's Blog","summary":"\u003ch2 id=\"me\" class=\"relative group\"\u003eMe \u003cspan class=\"absolute top-0 w-6 transition-opacity opacity-0 -start-6 not-prose group-hover:opacity-100\"\u003e\u003ca class=\"group-hover:text-primary-300 dark:group-hover:text-neutral-700\" style=\"text-decoration-line: none !important;\" href=\"#me\" aria-label=\"Anchor\"\u003e#\u003c/a\u003e\u003c/span\u003e\u003c/h2\u003e\u003cp\u003eHi there! I\u0026rsquo;m Abhishek, a tech writer and a journalist based in India. I am passionate about computers and technology.\u003c/p\u003e","title":"About"},{"content":"","date":null,"permalink":"https://kumarwrites.dev/topics/","section":"topics","summary":"","title":"topics"},{"content":"My thoughts, ideas, and experiences with the things I find interesting\n","date":null,"permalink":"https://kumarwrites.dev/writings/","section":"Writings","summary":"\u003cp\u003eMy thoughts, ideas, and experiences with the things I find interesting\u003c/p\u003e","title":"Writings"}]