Changelog

Best Skills repository pages have a new design. The install command sits right next to the description with a one-click copy button and tabs for the Skills CLI, Claude Code and plugins; the key numbers (stars, skills, forks, last update) are visible at a glance; long skill descriptions fold behind "Read more"; and repositories with many skills get a filter. Editor's Picks are marked with a highlighted note, and "Made for" lists the agents each repository says it targets.

Best Skills cards now carry highlight badges based on measured data: Trending (among the 20 fastest-rising repositories with 25+ new stars), the number of skills found in the repository, New (created in the last 30 days), Active (updated in the last 14 days) and Multi-agent (targets three or more agents). Hover a badge, or open "How the ranking works", to see what it means.

Every repository on Best Skills now has its own page — select a repository's name to open it. The page lists the skills inside the repository with each author's own description of what the skill does and when to use it, taken from its SKILL.md, plus install commands for the Skills CLI and Claude Code, where it ranks, key facts (stars, forks, language, license, last update) and similar skills. We collect skill lists once a day, so repositories fill in gradually.

Editor's Picks now always sit at the top of every ranking — Rising This Month, Best of the year, All-Time Top and New Arrivals — marked with the award badge instead of a rank number, so the ranking below them stays honest. Each tab says so in its description, and with an agent or type filter on, a pick stays on top only if it matches the filter.

Best Skills now loads more repositories automatically as you scroll, instead of a "Show more" button, and the counters show the real size of each ranking — for example 2,181 repositories in Best of 2026 and 2,532 in All-Time Top, where every tab used to stop at 100. Agent and type filters and search scroll the same way.

You can now submit your own skill to Best Skills. Open "Submit your skill" on the Best Skills page, paste the GitHub repository and add a short note; we check that it is a public skill repository — not a fork or an archive — and list it, with no minimum star count. You can follow the review under "Your submissions", and if your email address is verified we email you the result. You need a DevSly account, and each account can send up to 5 repositories a day. Listing is not a security review.

Logs now have a fixed retention period. Security, API, tool-usage and sign-in records are deleted automatically after two years by a daily clean-up job, and short-lived records such as proxy rate-limit entries and expired sign-in tokens much sooner. Webhook captures and load-test reports stay until you delete them. The Privacy Policy lists every period.

Load tests no longer keep the login details you enter for the target site — username and password, bearer token, API key or OAuth token. They are not saved with the test and no longer remembered in your browser's saved form data, and details kept by earlier tests were removed.

IP location lookups now run entirely on our own servers using the DB-IP Lite database instead of being forwarded to an outside geolocation service, so neither your IP address nor the addresses you look up leave DevSly, and lookups are faster. IP info results no longer include a region code or postal code, which this database does not provide. IP Geolocation by DB-IP (db-ip.com).

Google Analytics now loads only if you allow it. On your first visit a cookie banner offers Allow and Decline as equal choices, and declining removes any analytics cookies already set. You can change your choice at any time from Cookie settings in the footer.

You can now delete your account yourself under Account → Security. Confirm with your email address and current password; accounts created with GitHub can delete within 10 minutes of signing in with GitHub. Deleting removes the account together with its API keys, webhooks and their captured requests, load-test reports, connected AI assistants and usage history, and we email you a confirmation. Account deletions made by our team now remove the same linked data — before, some of it could stay behind.

The Terms of Service and Privacy Policy were rewritten from scratch so they describe what DevSly actually does: which tools run in your browser and which send data to our servers, what we log and why, and who we share data with. DevSly is operated by WAPI Solutions LLC; legal questions go to [email protected]. The Terms now include a security research section for penetration testers — scope, testing rules, reporting to [email protected], 90-day coordinated disclosure and safe harbour for good-faith testing — and /.well-known/security.txt points to it.

New: Best Skills — a ranking of AI agent skill repositories and awesome lists on GitHub, refreshed every day. Tabs show what is rising this month, the best of the year, the all-time leaders, new arrivals and hand-picked Editor's Picks, and you can filter by agent (Claude, Codex, Cursor, Gemini and more) or search. Every card shows stars, forks and recent growth and links straight to the repository. Rising This Month needs a month of daily history to be exact; until then it is an early estimate based on stars per day, and the page says so. Find it in the top bar next to MCP Tools.

The Gemini watermark remover was removing only one of the two marks on a Veo clip. Veo exports carry the sparkle and a "Veo" wordmark down in the corner; the tool repaired the sparkle and left the wordmark in every frame, so a video that looked finished still went out watermarked. It no longer assumes there is one mark. The watermarks are measured from your own video rather than looked up in a table, including the dark shadow around a mark, which used to fall outside the repaired area and leave a halo behind. Every mark found is repaired in the same single encode, so nothing is left for you to clean up in a second pass and no quality is lost to one. The upload page also no longer promises to trim an end card on Gemini clips, which do not have one.

A watermark that is only on screen for part of a clip is now only repaired for that part. The Gemini sparkle is typically lit for the opening stretch and gone afterwards, but the repair was being measured across the whole video and then applied to every frame of it. That did damage at both ends: the mark was only partly removed while it was up, and a faint dark ghost of it was stamped onto the footage after it had left. The tool now finds the stretch each mark is actually on screen, measures it there, and leaves every other frame exactly as you sent it.

The GEO score is now built on a written measurement contract instead of thirty-three weights scattered through the code. Three checks were quietly handing out free points: a file with no links scored full marks for "all 0 titles are meaningful" and "all 0 URLs are distinct", a site with no homepage title scored full marks for not repeating it, and a file with no headings at all scored full marks for being perfectly balanced. An empty set now earns nothing. Adding a short description to a link could previously LOWER your score, because depth was measured across described links only; it is now measured per link, so any description you add helps. The map-versus-guide decision was a cliff at exactly 80 links: one more line changed how four checks were graded. It is now a smooth blend, so nothing jumps. The letter grade is back, and grade, wording and card colour all come from one threshold table instead of three that disagreed. Weights now add up to exactly 100, with content worth half, and the page shows which checks could not be measured on your site rather than hiding a shrinking denominator. Two checks that had fallen out of the rubric are accounted for again. Locked in by a permanent regression suite of nine files and roughly 2,800 assertions, including frozen copies of real sites.

The score was judging every llms.txt as though there were only one right way to write one. A file that maps a whole documentation set — hundreds of links, every one pointing at a clean markdown copy — was penalised four separate times for its size, including by a words-per-link measure it could never satisfy. That is a measurement error, not a matter of taste, and it pushed genuinely good files into the fifties. The checker now recognises the two shapes a file can take, a curated guide and an exhaustive map, and judges each on its own terms. A summary written as an opening paragraph rather than a blockquote now counts too, with the difference named rather than scored as absent.

The GEO score now separates good files from average ones. Almost every site was landing in the seventies because a quarter of the score went to checks that nearly everyone passed, and because partial results were rounded into three buckets — 78% scored the same as 95%. Points that everyone earned have been removed, the weight moved to what a file actually says, and continuous measurements are now scored proportionally. Two new content checks ask how much is written per link and whether descriptions add anything to the titles. The letter grade is gone. Recommendations quote the offending line from your own file with its line number, and show the corrected version alongside it.

Long domain names were cut off on the GEO score card: the type size was picked from the character count, which cannot know how wide the text actually renders. The card now measures the name and shrinks to fit, down to a floor, and only truncates when even that will not do. Score cards also carry a version in their address now, so a fix to the artwork reaches you immediately instead of sitting behind a week of browser caching. Host names are lowercased on the way in, so the same site cannot appear twice.

LLMS Reader findings now tell you how, not just what. Every item opens into the steps for that specific fix — where the change lives, what to do in order, and a before/after snippet you can copy. Rows expand and collapse with an animation, on the saved report page and in the tool right after a scan. Because the steps follow the findings, and the findings differ per site, two reports no longer read like the same page with different numbers.

Every LLMS Reader report now reads like it was written about that site, because it was. The summary, the standout finding and the ranking against other scanned sites are generated from that site's own measurements — how many of its links point at markdown copies, whether its robots.txt blocks the assistants it just published a map for, how long since the file changed. The score itself is now the largest thing on the page, with strengths and gaps split into clearly coloured columns, and the same wording appears in the tool right after a scan.

The LLMS Reader no longer scores a site that has no llms.txt. Grading a file that does not exist read like a weak file rather than a missing one, so a site without the file now gets a plain answer instead: what a search engine reads from it today, a warning if robots.txt blocks AI crawlers, and a starter file already filled in with its own title and description. The latest-scores rail on the homepage now scrolls, with a pause control that also respects reduced-motion settings.

LLMS Reader scans are now permanent pages. Every scored site gets its own address at /network/llms-txt-reader/<site> with the full breakdown rendered server-side, a shareable PNG score card at /llms-score/<site>.png carrying the site's own favicon, and structured data so search engines can read both. The cards are declared in a dedicated image sitemap, and the latest scores appear as a rail on the homepage above the footer and on the tool page itself — in the HTML, not built by JavaScript afterwards.

New tool — LLMS Reader: type a site name, read its llms.txt and get a GEO score for it. It tries the standard path first, then the www and docs subdomains, and verifies the response is really a text file (a parked domain answering HTTP 200 with HTML is reported as a miss, not rendered as if it were real). Thirty-two checks across discovery, structure, link quality, search-engine basics and depth score the file out of 100 — including whether robots.txt blocks the very assistants the file was written for, whether the link list stays in the 10–40 range where it is still useful, and whether the homepage title just repeats your brand. A group that cannot be measured is dropped from the total rather than scored zero, and a site we cannot reach gets an explanation instead of a grade. Sites without an llms.txt are still scored on what a crawler reads today and get a starter file to copy. Every scan produces a shareable PNG score card, and the recent ones scroll in a gallery at the bottom of the page.

Replaced the inaccurate "Open Source" badge with "100% Free" and "Open API" — our REST API and MCP server are publicly documented, the site code is not published.

Layout fixes: footer tool links no longer stack in a single column, the plan badge on the account card is readable and centred again, and the account menu in the header is compact and simplified.

Load test report pages redesigned with clearer percentile bars (slow tails highlighted) and a print-friendly layout for sharing.

Subscription page rebuilt: a usage dial with colour thresholds, comparable plan cards, and a countdown to the daily quota reset.

Homepage tool ordering now puts featured tools first, then new ones — and your starred favourites always float to the top, in every sort mode.

Usage statistics never loaded: the page was requesting a stale address that no longer exists. Fixed across statistics, recent activity, export and clear.

Load test reports on the dashboard now have a "Load more" button instead of sending you to a separate page, and you can delete a report straight from the table.

New tool usage table on the dashboard — see your most used tools over 7, 30 or 90 days with run counts, average duration, a share bar and CSV export.

Account management moved into the dashboard: edit your profile or change your password in a quick modal without leaving the page, with instant feedback and no full page reload.

My IP page redesigned to a compact layout, and the third-party chat widget was removed for faster page loads.

A slim, theme-integrated LinkedIn bar now sits at the top of tool pages — reach us directly from any tool.

Header polished to a pixel-perfect rhythm (uniform 36px controls, tighter logo and spacing), and the Tools mega menu is now an opaque centered panel that dims the page behind it for clear reading.

Homepage: "This Week's Top Picks" and "All Developer Tools" merged into a single Tools Hub — a swipeable top-picks rail, compact filter pills and a denser row-card grid that fits more tools per screen. Duplicate tool cards in the grid were fixed.

API & MCP capability tags on every tool page: the breadcrumb now shows an "API" tag on tools with a REST endpoint and an "MCP" tag on tools available through the MCP server — hover either tag for a one-line usage hint with the endpoint address.

All ~70 tool pages now share one corporate skeleton: every tool was migrated to the shared template layer — consistent breadcrumb, database-driven page header and styling, with a single point of update for future design changes.

JavaScript Escape: fixed a bug where an example string terminated the page script early and dumped raw code as text; the reference table and guide moved to full width, the options became a compact segmented bar, and missing translations were added.

Tool pages: a new shared template layer renders breadcrumbs, headers and page metadata straight from the tools database (piloted on DNS Lookup, Base64 and UUID Generator).

DevSly is now on LinkedIn — follow links in the footer, the Quick Tools panel and the Ecosystem section for new tools and updates.

Homepage: the MCP and Chrome Extensions sections merged into a single "DevSly Ecosystem" block, the hero is leaner with live stats, and roughly 2,000 lines of dead CSS were removed for faster loads.

Quick Tools panel rebuilt as a command palette: search-first layout, category filter chips, a swipeable quick-access slider and a category-grouped tool list. Open it with Cmd+K / Ctrl+K.

MCP quick-access topbar: the MCP endpoint now sits above the header with one-click copy, next to a direct link to the API documentation.

Corporate design refresh across the whole site: a brand-blue design system aligned with our logo colors, a compact 60px header, a theme-aware footer and shared design tokens — consistent in light and dark mode.

New tool: AI Humanizer — rewrites AI-generated text into natural prose with a 300+ rule engine (clichés, LLM vocabulary, wordiness, typography, contractions) for English and Turkish, and scores the AI signature across 12 heuristic signals. Runs fully client-side.

The MCP connection guide moved into the API documentation (/api/docs) with an expanded tool reference and OAuth details, and the homepage now has a section for connecting DevSly to your AI assistant.

Acknowledgement emails and certificates now reflect the points system: certificates show the points awarded, and emails show this report’s points, your season total, and your current rank — with a dedicated "new report added" email when you climb the board again.

Security Hall of Fame is now a points-based leaderboard. Each accepted report earns severity-weighted points (Critical 40, High 20, Medium 10, Low 5, Info 1), and researchers are ranked per year with a top-3 podium, an All-time view, and a per-researcher report ledger. The board resets every January.

The MCP server is protected by a full OAuth 2.1 authorization server (PKCE, exact redirect-URI matching, short-lived tokens with refresh rotation, per-user rate limits). Network tools stay SSRF-guarded, uploads are validated by content signature and stored web-inaccessibly, and data captured by webhooks is treated as untrusted.

Dashboard: new "AI / MCP" tab. See the MCP connection URL, the full list of AI-accessible tools, and every AI app connected to your account — with one-click revoke to cut off access anytime.

New: DevSly MCP server. Connect DevSly to Claude — or any AI assistant that supports remote MCP — and run your developer tools straight from the chat: IP/DNS/HTTP lookups, the webhook tester, and the NotebookLM & Gemini video watermark remover. Add https://devsly.io/mcp in your AI client; sign-in uses secure OAuth (no API key to paste). See the setup guide at /mcp-connect.

Hardened post-logout browsing: pages opened while signed in now force a fresh check when reached via the browser Back button, so a stale signed-in view can no longer be shown after logout. (Sessions were already destroyed server-side and protected pages already re-checked the session — this closes the cached-view gap.)

The X button on the login/register popup now closes it. A decorative animated layer in the popup header was sitting on top of the button and swallowing the click.

Gemini Watermark Remover now removes the sparkle correctly. Watermark detection is tuned per source: the method built for NotebookLM's text mark was misreading Gemini clips (live footage with a static camera) and repairing the wrong part of the frame. Gemini now uses the sparkle's known shape, so only the mark is repaired and the picture behind it is preserved.

New: Forgot Password. You can now reset a forgotten password from the login screen — we email a single-use link that expires after 1 hour. Accounts created with GitHub keep signing in through GitHub.

Video tools: the cleaned file no longer downloads automatically — a "Download Cleaned Video" button appears when processing finishes. Added a staged progress indicator (Upload → Process → Ready) and a refreshed, animated interface.

Video watermark removal is now far more precise: the watermark is auto-detected in every video instead of assuming a fixed position, and only the logo's own pixels are repaired (shape mask) instead of blurring a whole rectangle. This fixes marks that were partially left behind on letterboxed videos and keeps artwork underneath the watermark intact.

New tool: Gemini Watermark Remover — remove the Gemini (Veo) sparkle watermark from every frame of an AI-generated MP4. Free, up to 10 videos/day; files are processed then deleted, never stored.

New tool: NotebookLM Video Watermark Remover — remove the NotebookLM watermark from every frame of an MP4 and automatically trim the end card. Free, up to 10 videos/day; files are processed then deleted, never stored.

New tool: NotebookLM Watermark Remover — remove the NotebookLM watermark from PDF slide exports for free, with no upload and no sign-up. Runs fully client-side in your browser.

cURL Converter: new "Send" button runs the request entirely in your browser (your IP) and shows the live response — our server makes no outbound requests. Works for CORS-enabled APIs; clear guidance shown when a target blocks cross-origin browser requests.

New tool: cURL Converter — convert any cURL command to request code in 35+ languages and frameworks (Python, PHP, JavaScript, Node.js, Go, Java, C#, Ruby, Rust, Swift and more). Runs fully client-side.

API Documentation - Updated docs with new mode parameter, bulk endpoint examples, and meta response fields

WHOIS API Performance - Reduced timeout from 10s to 5s, parallel execution for WHOIS + Dynadot queries

Domain Cache System - Smart file-based caching with TTL (available: 5min, registered: 1hr, pricing: 1hr)

RDAP-Only TLD Support - Automatic detection for Google TLDs (.dev, .app, .page) that dont support traditional WHOIS

Bulk Domain Check Endpoint - Check up to 100 domains in a single request at /api/v1/network/domains/check

Fast Mode for WHOIS API - New mode parameter (fast/full/availability_only) for 3x faster domain lookups (~500ms vs 2-3s)

Domain Availability System - Professional domain checking system with real-time pricing and availability detection for 810+ TLDs

Simplified XenForo Password Generator Developer Guide - removed Examples and Integration sections, keeping only essential Overview information

Updated XenForo tool URL structure to /hash/xenforo-password-generator/ for better SEO and clarity

Added routing /hash/xenforo/ for XenForo Password Generator tool

XenForo tool: Integration examples for forum migration, SSO, and API authentication

XenForo tool: Complete developer guide with PHP, Python, Java, and Node.js examples

XenForo tool: Password verification against XenForo hashes (serialized or raw)

XenForo tool: Batch processing for multiple passwords with CSV/JSON export

XenForo tool: Supports both raw bcrypt and PHP serialized format output

Added XenForo 2 Password Hash Generator tool - Generate XenForo compatible bcrypt password hashes

Fixed inconsistent padding and spacing across trending cards for balanced layout

Fixed ranking badges visibility - all badges now fully visible with proper dimensions

Fixed trending section layout issues where content was overflowing boxes

Enhanced mobile responsiveness with proportional scaling while maintaining professional appearance

Removed Weekly Summary section with fake total growth data

Removed fake statistics (growth percentages, user counts) for more authentic presentation

Optimized card proportions with fixed icon sizes (64px box, 48px icon) for consistent visual hierarchy

Implemented three-section card layout: top bar (ranking + badge), main content (icon + details), action footer

Added color-coded status labels (Most Used, Popular, Hot, Featured, Top Pick, Trending) with specific color schemes

Enhanced ranking badges with gold, silver, and bronze gradient designs for top 3 positions

Redesigned trending section with balanced, professional corporate layout

Modified user_webhooks table to support NULL user_id for anonymous webhook creation

Fixed missing FontAwesome icons for My Webhooks menu, HTTP Headers tool, and Webhook Tester tool

Added URL parameter support to Webhook Tester for viewing specific webhooks via webhook_id parameter

Completely redesigned Webhook Tester page with corporate design matching dashboard aesthetics, CSS variables, and minimal spacing

Completely redesigned Webhook History page with professional, dashboard-aligned design including horizontal stats bar, table layout, and dark mode support

Added "My Webhooks" menu item to user dropdown for easy access to webhook history

Created new session-based webhook API endpoints: /api/v1/network/my-webhooks, /api/v1/network/my-webhook-create, /api/v1/network/my-webhook-delete, /api/v1/network/my-webhook-requests

Added session-based authentication for webhook system, allowing both logged-in users and anonymous users to create webhooks

Fixed Apache Authorization header not being passed to PHP for API authentication

Fixed webhook tester 500 error caused by incorrect file ownership permissions

Added floating API Tester widget that appears on all pages after 3 seconds, with smart dismissal (reappears after 24 hours)

Added API Testing Suite featured section on homepage with live demo and feature highlights

Reorganized time tools into Developer Tools category for better discoverability

Added Date Formatter with 15+ format options including ISO 8601, RFC 2822, Unix timestamps, SQL formats, and custom formats with one-click copy

Added Date Difference Calculator with business days calculation, weekend tracking, and detailed breakdown in years, months, days, hours, minutes, and seconds

Added Timestamp Converter tool with real-time updates, supporting both seconds and milliseconds, with quick action buttons for common timestamps

Added risk assessment for each security check with detailed attack vectors and mitigation strategies

Comprehensive SEO-optimized content explaining security headers, implementation guides for Apache/Nginx, and best practices

Quick metrics dashboard showing response time, HTTP status, SSL verification, compression, and redirects

Export analysis results as JSON or CSV format

Interactive filtering by security status (all, failed, passed, critical, high)

Real-time analysis of HSTS, CSP, X-Frame-Options, CORS, and modern security headers (COEP, COOP, CORP)

Security score calculation with critical, high, medium, and low severity classifications

HTTP Security Headers Analyzer - Advanced security header analysis tool with instant grading (A-F), 13+ security checks, and detailed recommendations

Major Regex Builder upgrade: Added Pattern Library with 20+ predefined patterns, Performance Analyzer with complexity scoring, multi-language Code Generator (8 languages), Pattern History with localStorage, Find & Replace mode, Lookaround support, advanced Pattern Explanation engine, and enhanced UI with tabbed interface

Enhanced mobile menu design with modern UI, smooth animations, organized sections, and improved user experience

Fixed mobile menu not opening issue by implementing complete mobile navigation functionality with proper toggle, overlay, and responsive design

WHOIS: Fixed .de lookups by sending DENIC -T dn and adding DENIC-specific parsing. Nameservers and updated date now display correctly; optional expiry is captured when present via registrar proxy. Other TLD behavior remains unchanged.

Fixed mega menu CSS conflicts by adding specific .mega-menu prefixes to avoid style overlaps

Added Load Test Reports link to user dropdown menu in header for easy access

Implemented delete report functionality with confirmation dialog

Added CSV export functionality for load test reports with comprehensive metrics

Created detailed load test report page with performance metrics, percentiles, and interactive charts

Implemented Load Test Reports section in user dashboard showing last 5 test results

Added comprehensive Load Test Reports system with list view, detailed reports, and chart visualizations

Modernized tools-content layout in mega menu with responsive grid and card-based design

Cleaned up excessive z-index values in index.php reducing from 10000/9999 to reasonable levels (30-50)

Fixed tools-content appearing under tools-sidebar in header mega menu by converting to grid layout

Implemented load test report download functionality with detailed PDF generation including all test metrics

Added visual success rate progress bars with color coding based on performance thresholds

Fixed chart rendering issues on load test reports page with proper height constraints and initialization

Removed standalone login and register pages in favor of modal-based authentication

Fixed database column name mismatches in load test queries (percentile_*, results_data, max_response_time)

Fixed load test reports page session handling and login redirect issues

Fixed modal registration password confirmation field mismatch error

Redesigned dashboard load test reports section with modern card-based layout and interactive hover effects

Enhanced load test reports page design with success rate indicators, status animations, and improved metric cards

Added professional PDF export for load test reports with visual design including charts, progress bars, and color-coded metrics

Implemented comprehensive input validation and XSS protection for new tools

Fixed WebSocket API documentation visibility in API docs

Added SEO sections to Docker Compose Builder

Enhanced Dockerfile Generator with SEO content and professional design

Updated tool analysis report with 30+ new tool recommendations

Fixed icon visibility issues for WebSocket Tester, SHA384 Generator, and Dockerfile Generator

Added Docker Compose Builder with service dependency visualization

Added Kubernetes YAML Generator with visual configuration builder

API Testing Suite: Modern dark theme UI with animations

API Testing Suite: Enhanced security with XSS protection and URL validation

API Testing Suite: Environment variables and request parameters

API Testing Suite: Authentication support (Bearer, Basic, API Key)

API Testing Suite: Support for GET, POST, PUT, PATCH, DELETE methods

API Testing Suite: Collection management and request history

Added API Testing Suite - Professional API testing tool with multi-tab support

Added automatic random tool backfill to ensure 6 tools always display in trending section

Implemented database-driven trending tools section showing most used tools from last 30 days

Changed mega menu navigation from click to hover interaction for better UX

Implemented dynamic search results container with proper positioning and cleanup

Fixed hero section search functionality - results now display correctly in dropdown

Added WHOIS, My IP, HTML Editor, Sitemap Generator, JS Obfuscator, SQL Query Builder, QR Generator, and WordPress Hash to quick sections

Updated quick access tools in header with new tool selection

Fixed incorrect creation dates in WHOIS results

Fixed WHOIS nameserver display issue for .co, .de, .io and other TLDs

Added comprehensive date parsing for WHOIS results supporting 18+ date formats

Enhanced WHOIS tool with support for 90+ TLD servers and improved nameserver extraction

Added Social Media Preview tool to preview how links appear on Google, Facebook, Twitter/X, and LinkedIn

Standardized border-radius to 8px across all components for consistent visual language in the minimalist design system

Implemented smooth hover transitions (0.2s ease) across all interactive elements with translateY effects for better user feedback

Removed tool description paragraphs from All Tools cards in minimalist redesign, showing only tool names for cleaner visual hierarchy

Added proper error handling for database connection failures with fallback to static tool data, ensuring tools are always displayed even if database is unavailable

Simplified All Tools section heading from complex HTML with colored spans to clean "All Tools" text with "Simple. Fast. No BS." tagline

Added search functionality helper text indicating Ctrl+K keyboard shortcut for quick search access, improving discoverability of keyboard navigation features

Optimized JavaScript in All Tools section by removing unnecessary IntersectionObserver animations and setTimeout delays that were causing rendering issues

Fixed header.php Quick Tools navigation by removing erroneous /tools/ prefix from all quick access links after initially adding it incorrectly

Created minimalist changelog display page at /changelog.php with clean typography, color-coded update types, responsive design, and full dark mode support

Implemented comprehensive SQL-based changelog system with versioning support, categorized update types (feature/bugfix/improvement/security), and date-based grouping for better update tracking

Changed SHA256 Generator icon from generic fa-fingerprint to more appropriate fa-shield, better representing security and hash generation functionality

Updated Font Awesome icon references: changed JSON Formatter from non-existent fa-brackets-curly to fa-file-code, ensuring all tool icons render correctly

Implemented aggressive CSS override rules with !important declarations to force visibility of All Tools section, addressing display issues caused by conflicting stylesheets

Added comprehensive fallback data structure covering all 8 tool categories (Encoders, Hash Generators, Formatters, HTML Tools, JavaScript Tools, Network Tools, Misc Tools, Database Tools) with proper display names

Removed problematic t() translation function dependencies from trending tools section, replaced with direct string manipulation for tool names and descriptions

Fixed and optimized categories-container CSS with proper grid display, forced visibility rules, and removal of conflicting animation styles that were hiding content

Corrected all tool URLs in index.php from incorrect paths to proper structure (e.g., /tools/encoder/base64/ format) ensuring proper routing throughout the application

Resolved file permission issues by executing chown -R devsly:devsly command on public_html directory, fixing MIME type errors for CSS and JS files

Significantly improved dark mode implementation for All Tools section with proper contrast ratios, border adjustments, and hover state refinements for better accessibility

Fixed tool visibility issue in All Tools section by implementing display_name field mapping for database entries and comprehensive fallback data structure for all tool categories

Enhanced All Tools grid layout with responsive columns (auto-fill, minmax 280px), consistent card heights, improved spacing (16px gap), and mobile-first approach for better usability

Implemented real-time search functionality in All Tools section with instant filtering, category hiding when no matches found, and keyboard shortcut (Ctrl+K) for quick access to search input

Redesigned entire All Tools section with minimalist design philosophy: removed gradients, simplified color palette to monochrome, reduced visual noise, and implemented clean grid layout with subtle hover effects

Added Password Generator as the primary tool in Quick Access toolbar for improved security workflow, positioned as the first item with key icon for instant access

Fixed critical navigation issue in header Quick Tools where all links were pointing to incorrect URLs (e.g., /json-formatter instead of /formatter/json/beautify/) causing 404 errors

Completely removed Quick Switcher (Ctrl+P) functionality including all JavaScript event handlers, CSS animations, and DOM elements to improve page performance and reduce code complexity