What exactly is OryoNex?

OryoNex is an AI-powered work management platform with Kanban, List, Gantt, and Calendar views. It combines OryoAgent (a 120+ capability AI assistant), real-time collaboration, e-commerce tools, a customizable Widget Dashboard, and gamification in one workspace.

What's the difference from other Kanban tools?

Traditional tools list tasks. OryoNex adds OryoAgent with 120+ AI capabilities, browser automation, desktop control, social media content creation, multi-agent orchestration, and gamification — helping teams not just organize but execute intelligently.

What is a Workspace? How many can I create?

A workspace is a separate organizational area for each project. Each workspace has its own board, team members, tasks, and settings. The number of workspaces depends on platform settings and your plan, and can be increased by an admin if needed.

How do AI credits work? Why a credit system?

AI features consume credits. Credits belong to a user account; within a workspace, AI usage typically deducts from the workspace owner's (first admin) balance. The credit system helps manage AI usage and cost.

Is my data secure? Is there a backup system?

Data is stored in PostgreSQL. All communication is protected with TLS/HTTPS; RBAC is enforced in the app and chat messages are encrypted with AES-256-GCM. Backup processes are handled according to the operational plan. Only you and authorized team members can access your data.

Changelog - OryoNex Development Log

Changelog

Follow OryoNex's journey of growth. Every update is a step towards a better experience.

15 Nisan 2026v5.0

OryoAgent v5.0 — Smarter Planning and Unified Commands

OryoAgent now plans your tasks with greater precision and lets you combine multiple actions in a single command. Browser automation and task management work together seamlessly.

NewUnified commands — browse a website and create a task in one message (e.g. "go to amazon.com, list prices, then save as task")

ImprovedPlan quality — the agent now evaluates safety, accuracy, goals, and efficiency as layered priorities

ImprovedCommand understanding — Turkish and English natural language commands are interpreted more precisely with fewer misroutes

FixedRetina display support — browser window sizing corrected for high-resolution Mac screens

ImprovedOverall stability — consistency improvements in plan revision, observation, and response generation steps

10 Nisan 2026v4.57

Three New Agent Quick Actions — Web Fetch, Workspace Files, Step-by-Step Reasoning

OryoAgent quick action panel now exposes three additional capabilities directly from the Analytics → Research section. Each one is available as a one-click button in the chat sidebar or through natural language in Turkish and English.

NewFetch Web Page — pull clean readable content from any URL or extract specific elements by CSS selector for quick page scraping

NewWorkspace Files — read, write, and organize files in your workspace's isolated storage area. Useful for saving intermediate reports and persistent notes

NewStep-by-Step Reasoning — ask the agent to break a complex problem into numbered reasoning steps and produce a structured conclusion

ImprovedQuick action panel — Analytics → Research section now lists the full set of five MCP-powered capabilities alongside existing research tools

9 Nisan 2026v4.56

MCP Integration — Extend Your Agent with External Tools

OryoAgent can now connect to external tool servers using the Model Context Protocol (MCP) standard. Parse documents, fetch web pages, and build a knowledge graph — all from your agent chat. Workspace admins can add custom MCP servers for even more capabilities.

NewDocument parsing — upload and analyze PDF, Excel, Word, and CSV files directly in agent conversations

NewWeb page reading — fetch any web page and get clean, readable content extracted automatically

NewKnowledge graph memory — the agent can now store and recall entity relationships (customers, products, suppliers) across conversations

NewWorkspace file access — the agent can read, write, and organize files in your isolated workspace storage area

NewStep-by-step reasoning — complex problems can be broken into structured thinking steps for clearer answers

NewCustom MCP servers — workspace admins can connect additional tool servers from the Integrations settings

ImprovedWorkspace integrations — new MCP server management section with connection testing, status monitoring, and easy configuration

ImprovedAccessibility — improved form labels, keyboard navigation, and screen reader support throughout integrations settings

9 Nisan 2026v4.55

Chat History Search — Find Any Message Instantly

Full-text search across all your agent conversations. Type in the search bar and find any message by content — results are grouped by conversation with highlighted snippets. Click a result to jump directly to that message.

NewFull-text message search — find any past conversation by typing keywords in the search bar

NewHighlighted snippets — matching text is highlighted in search results for quick scanning

NewJump to message — click a search result to open the conversation and scroll directly to the matching message

ImprovedSearch experience — inline loading spinner, clear button, minimum character hint, and error feedback

ImprovedAccessibility — proper ARIA labels, keyboard navigation support, and consistent focus indicators throughout search

6 Nisan 2026v4.54

OryoAgent — Smarter Browser Automation & Performance Boost

OryoAgent's browser automation is now significantly faster and more accurate. Inspired by modern accessibility-first approaches, the AI now understands element states (open/closed menus, checked boxes, active tabs) and form labels — leading to fewer misclicks and smarter navigation. Performance improvements cut task completion time by up to 6x.

NewElement state awareness — the AI now detects whether menus are open or closed, checkboxes are checked, and which tab is active

NewForm label detection — input fields are now identified by their associated labels, improving form-filling accuracy

NewPopup menu hints — the AI recognizes buttons that open dropdown menus and uses hover-first strategy

ImprovedBrowser session continuity — shopping cart and filter states are now preserved across multi-step tasks

ImprovedBrowser window now opens in full screen automatically for better page visibility and navigation accuracy

FixedBrowser automation speed — tasks that previously took over 3 minutes now complete in under 40 seconds

FixedReduced unnecessary waiting during page verification steps for faster task completion

2 Nisan 2026v4.53

OryoAgent — Smart Memory: User Profile, Decision Log & Self-Improvement

OryoAgent now learns your preferences, remembers decisions, and improves itself weekly. Inspired by the latest AI agent research (Mem0, PersonaMem-v2, Reflexion pattern), this update adds three major intelligence layers that compound over time — the more you use OryoAgent, the smarter it gets.

NewUser communication profile — the AI learns your preferred response style, technical depth, and format over time

NewImplicit preference learning — each successful interaction refines how the AI understands your work style

NewAutomatic decision log — key decisions are extracted from daily conversations and stored in memory

NewContradiction detection — the AI remembers past decisions and alerts you when a new request conflicts

NewWeekly self-improvement engine — analyzes usage patterns, failure rates, and recipe effectiveness

NewProactive weekly suggestions — the AI recommends specific optimizations based on your workspace activity

NewPattern detection — frequently used capabilities are identified and tracked for workflow optimization

NewRecipe health monitoring — recipes with high failure rates are automatically flagged for review

ImprovedAI memory now includes a rich user profile section, recent decisions, and workspace usage patterns in every response

ImprovedMemory system upgraded with better search, 30-day decision retention, and automatic cleanup

ImprovedAll learning features only activate after verified successful interactions for accuracy

ImprovedCron notification system now delivers alerts via real-time WebSocket for instant feedback

1 Nisan 2026v4.52

Space Invaders — Gameplay & Reliability Update

Major gameplay improvements to Space Invaders with new scoring mechanics, visual effects, and critical stability fixes for the AI assistant.

NewStyle Rank system — earn D/C/B/A/S/SS ranks based on wave performance

NewPerfect Wave bonus — flawless play rewards extra points with a golden banner

NewWave clear bullet cancel — remaining enemy projectiles convert to score gems

NewPower-up expiry warning — blinking indicator and timer bar when time is running out

ImprovedBomber bombs now have a distinct visual with ground-level blast radius warning

ImprovedHit effects redesigned with shape-matching animations

ImprovedWave transition screen now shows accurate scaled bonus amounts and wave score

FixedFixed a critical gameplay issue where the player could become permanently invulnerable

FixedFixed bomber special attacks not triggering properly at ground level

FixedFixed scoring information display showing incorrect point values

FixedAI assistant no longer gets blocked by stale background processes

31 Mart 2026v4.51

Widget Dashboard — Settings, Weather & Quality Update

Widgets now have customizable settings. New weather widget with current conditions and 5-day forecast for 10 cities. All widget layouts redesigned, empty state indicators added across the board.

NewWidget settings — customize period, city, currency, and display preferences per widget

NewWeather widget — real-time conditions and 5-day forecast for 10 cities

New32 total widgets for an enriched dashboard experience

ImprovedConsistent widget card sizing and layout across all presets

ImprovedInformative empty state indicators for widgets without data

ImprovedMeeting Calendar widget now shows real task deadlines

FixedResolved widget save error when adding new widgets

FixedFixed duplicate widget header display issue

31 Mart 2026v4.50

Widget Dashboard — Customizable Dashboard Panels

5 industry-specific presets, 30 drag-and-drop widgets, real-time data connections, and OryoAgent-powered widget management. Custom panels for e-commerce, developers, business management, marketing, and freelancers.

NewWidget Dashboard — fully customizable drag-and-drop dashboard

New30 different widgets: charts, metrics, calendars, lists, and timers

New5 industry presets: E-commerce, Developer, Business, Marketing, Freelancer

NewReal-time data from Google integrations: SEO score, traffic sources, campaign performance

NewOryoAgent widget manager — add, remove, or suggest widgets from chat

NewSmart data caching — widgets auto-update when agent performs analysis

NewThree widget states: empty, error, and stale data indicators

ImprovedWorkspace-level isolation — each workspace sees its own widget data

ImprovedMobile-responsive layout — single column on small screens

ImprovedModern dark theme widget designs with consistent visual language

29 Mart 2026v4.49

OryoAgent — Read Web Page Capability

OryoAgent can now read and extract content from any public web page. Just paste a URL and get clean, readable text — no browser automation needed.

NewNew 'Read Web Page' capability — extract content from any URL as clean text

NewAutomatic fallback for protected sites ensures maximum content accessibility

NewCSS selector support to extract specific sections of a page

NewZero credit cost — page reading is completely free

ImprovedURLs without https:// prefix are now automatically corrected

29 Mart 2026v4.48

Space Invaders — Performance & Gameplay Improvements

13 improvements for a smoother, fairer, and more performant Space Invaders experience. Dive attack game-over fixed, power-ups now preserved during pause, and mobile controls refined.

FixedDiving enemies no longer trigger unfair instant game over

FixedPower-ups no longer expire while the game is paused

FixedImproved mobile touch control accuracy when sliding fingers between zones

FixedVisual flickering on diving enemies resolved

ImprovedBullet and effect rendering performance optimized — higher FPS

ImprovedDynamic adaptation to screen size changes added

29 Mart 2026v4.47

Space Invaders — Shield Stacking & Extra Life System

New power-up mechanics bring deeper strategy to Space Invaders: collect shields consecutively to stack extra firepower, pick up extra lives during boss fights, and enjoy an enhanced drop system.

NewExtra Life power-up — collectible health drops now appear during gameplay and boss encounters

NewShield Stacking — collecting shields consecutively adds extra bullets in a fan spread pattern, up to 5 simultaneous shots

NewFirst boss guaranteed life drop — defeating the first boss phase rewards an extra life

ImprovedShield drops are now more frequent, making the stacking mechanic more accessible

ImprovedMaximum lives increased from 4 to 5 to complement the new extra life system

ImprovedShield visual feedback — stacked shields display expanding pulse rings around the player ship

28 Mart 2026v4.46

AI Agent Capability Expansion — 13 New Tools + 11 Enhanced

OryoAgent now features 112 capabilities across 11 categories, adding product management, customer service, team operations, and marketing intelligence tools alongside major enhancements to existing e-commerce capabilities.

NewBrainstorm tool — structured idea generation with categorization, scoring, and actionable next steps

NewPRD Generator — create product requirements documents with user stories, acceptance criteria, and technical specifications

NewLaunch Planner — build phased product/campaign launch plans with milestone tracking and task creation

NewFeature Prioritizer — prioritize features using RICE, MoSCoW, and custom scoring frameworks

NewDeal Tracker — sales pipeline analysis with opportunity tracking, forecasting, and risk assessment

NewReview Responder — generate professional responses to customer reviews across multiple tones and platforms

NewFAQ Generator — create comprehensive FAQ lists for products and services with SEO-ready schema markup

NewChurn Analyzer — customer churn risk analysis with retention strategies and win-back campaign suggestions

NewSupport Ticket Classifier — categorize support tickets by type, priority, and suggested response

NewTeam Announcement Writer — draft professional team communications for updates, launches, and milestones

NewWorkflow Automator — analyze business processes and suggest automation opportunities with flow designs

NewBundle Recommender — product bundle and cross-sell recommendations based on catalog analysis

NewInfluencer Finder — discover relevant influencers for your target audience with outreach strategy

Improved11 existing e-commerce tools enhanced with deeper analysis, richer outputs, and broader coverage

ImprovedExpanded intent routing with 333 new test cases ensuring accurate capability detection in both Turkish and English

27 Mart 2026v4.45

Space Invaders — Major Game Overhaul

A complete transformation of the Space Invaders mini-game with 5 unique boss encounters, 15 waves, new enemy types, advanced power-ups, and arcade-inspired mechanics from iconic shooters.

New5 unique bosses with distinct visuals, 3-phase battles, and signature attack patterns across waves 3, 5, 7, 10, and 15

New3 new enemy types: armored tanks with aimed shots, fast-moving swarms with zig-zag patterns, and bombers with area-denial attacks

New3 new power-ups: continuous laser beam, homing missiles that track enemies, and time slow that reduces enemy speed

NewGrazing mechanic — earn bonus points when enemy bullets pass dangerously close, inspired by classic bullet-hell games

NewDash dodge with brief invincibility — press Shift + direction for a quick evasive maneuver with afterimage trail

NewMulti-kill slow motion — rapid consecutive kills trigger a brief dramatic slowdown effect

NewDive attacks — enemies break formation and swoop toward you in curved paths, inspired by Galaga

ImprovedExtended from 10 to 15 waves with a saw-tooth difficulty curve — breather waves after boss fights prevent fatigue

Improved5 difficulty tiers: Easy, Medium, Hard, Hell, and the new Nightmare tier for the final waves

ImprovedBoss preview warnings between waves — see the next boss's name and power level before the battle begins

ImprovedRicher end-game stats: enemies killed, bosses defeated, UFOs shot, max combo, playtime, and wave progress

ImprovedWeighted power-up rarity system — common, uncommon, rare, and very rare drops for strategic variety

NewHitstop freeze frame on boss kills and chain milestones — dramatic impact feel inspired by fighting games

NewChain kill system with milestone callouts: AMAZING at 10 chain, UNSTOPPABLE at 25, LEGENDARY at 50 — with score bonuses

NewPoint-blank bonus — close-range kills earn 3x score multiplier with visual callout

NewBullet cancel on boss phase transitions — all bullets convert to score gems, plus 3-4 power-up drops rain down

NewBoss rage burst system — periodic intense attack windows with faster bullets for unpredictable challenge

ImprovedComplete mobile touch redesign: 3-zone layout (move left / fire / move right) with visual indicators and dedicated dash button

ImprovedHaptic feedback on mobile: vibration patterns for shooting, hits, boss encounters, and power-up pickups

ImprovedAll UI buttons upgraded to 44px+ touch targets with proper touch event handlers for mobile play

27 Mart 2026v4.44

Smart Quick Actions Panel — Organized with Subcategories

The agent quick actions panel now groups capabilities into clear subcategories within each main category. E-Commerce's 38 actions are organized into 6 focused groups, Project Management into 4, and Analytics into 4 — making it easier to find the right tool at a glance.

NewSubcategory grouping in quick actions: E-Commerce split into Product & Catalog, SEO & Content, Ads & Marketing, Sales & Analytics, Stock & Data, Pricing & Promotions

NewProject Management grouped into Task Actions, Planning & Analysis, Reporting, and Scheduled Tasks

NewAnalytics organized into Research, Reporting, Team & Workflow, and Developer sections

ImprovedMode selector tooltips for X Content Creator — hover to see what each mode does

ImprovedMobile keyboard now shows standard text input for content creation instead of URL keyboard

27 Mart 2026v4.43.1

X Content Creator — Security & Quality Hardening

Deep review and hardening of the X Content Creator capability with 26 fixes across security, prompt quality, output reliability, and user experience.

ImprovedEnhanced content safety with advanced prompt protection and input validation

ImprovedStructured AI output enforcement for all 5 creator modes — more consistent and reliable results

ImprovedBetter Turkish content quality with proper character handling and brand name preservation

ImprovedCharacter limit enforcement now covers all modes including Polish and Reply

FixedStandalone 'X' keyword now correctly routes to X Content Creator

27 Mart 2026v4.43

X Content Creator — AI-Powered Tweet & Thread Generation

OryoAgent now helps you create X (Twitter) content with 5 specialized modes. Generate tweets, build threads, improve existing posts, convert blog content, and craft strategic replies — all personalized to your voice through style DNA analysis.

NewX Content Creator with 5 modes: Brainstorm (3 versions), Polish, Thread Builder, Content Repurpose, and Reply Strategy

NewAlgorithm-aware content generation with engagement signal predictions per version

NewPersonal voice preservation through Mirror DNA — your writing style is remembered and reflected in every generation

NewCombined input modal with mode quick-select buttons for seamless content creation workflow

NewSmart follow-up actions: add as kanban task, expand to thread, or generate alternatives with different hooks

26 Mart 2026v4.42

Space Invaders — New Game in Game Center

Space Invaders has been added to the Game Center. A classic arcade experience with 10 waves, 4 enemy types, boss fights, power-ups, and a barrier system. Each game now has its own leaderboard with a game selector.

NewSpace Invaders game — 10 waves, progressive difficulty, wave transition animations, and victory screen

New4 enemy types: Bug, Warning, Task (workspace integration) and UFO (bonus) with space-themed arcade gameplay

New2 boss fights — Mini Boss (Wave 5) and Final Boss (Wave 10) with entrance animations and HP bars

New4 power-ups: Rapid Fire, Shield, Double Shot, and Bomb — randomly dropped by defeated enemies

NewPer-game leaderboard selector in Game Center — each game has its own weekly and all-time rankings

NewHidden easter egg: Enter a special key combination on any page to launch a quick overlay game session

ImprovedParallax starfield background, enemy hit effects, score popups, boss red vignette, and combo screen flashes for a rich visual experience

ImprovedGame Center cards redesigned — active and coming soon games are now visually distinct

ImprovedDifferent sound effects per enemy type and a dramatic boss entrance sound added

25 Mart 2026v4.41

GEO & Site Audit — Real Page Analysis

GEO Report and Site Audit now fetch and analyze actual page content. When server access is blocked, automatic local browser fallback via MacBook kicks in.

ImprovedGEO Report now fetches actual page HTML for analysis — heading hierarchy, FAQ sections, Schema.org markup, and word count are measured from real page data

ImprovedAutomatic local browser fallback when server access returns 403 (for both Site Audit and GEO Report)

ImprovedRedesigned input modal for GEO and Site Audit — URL input and analysis type selection are now combined in one screen

ImprovedAnalysis type parameters are automatically recognized when sent as Turkish text (e.g. "tam analiz" → full, "teknik kontrol" → technical)

ImprovedGEO report output now shows Schema.org types and data source (server/local browser) information

25 Mart 2026v4.40b

E-Commerce Operations Assistant

OryoAgent now automates daily e-commerce operations. 12 new capabilities added including stock tracking, price research, content quality checks, and campaign validation.

NewStock Comparison — compares two files by barcode and generates an update file (saves 15-20 minutes daily)

NewOut of Stock Detection — automatically finds products that are out of stock but still displayed in the storefront

NewMissing Content Detection — identifies missing product data fields and calculates a quality score

NewDaily Sales Summary — compares today's sales with yesterday and last week, detects anomalies

NewAlt Text Generator — produces SEO-friendly alternative text and file name suggestions for product images

NewSchema Generator — creates search engine-friendly structured data (JSON-LD) for product pages

NewAgency Performance Report — reports external agency task delivery rates and delay analysis

NewPrice Research — researches market prices for any product across multiple platforms

NewCategory Suggester — suggests appropriate e-commerce categories based on product information

NewCoupon Code Generator — generates unique coupon codes and creates bulk import files

NewMail Triage — automatically classifies operational emails and creates task lists

NewCampaign Validation — validates campaign rules, generates test scenarios and risk analysis

ImprovedLarge file support — increased file reading limits from local device, enabling processing of data files with thousands of rows

22 Mart 2026v4.33

Google Services Integration

Connect all Google APIs with a single service account. OryoAgent now analyzes and reports using real data.

NewGoogle Services integration type — access Search Console, Analytics, Ads, and PageSpeed data with one service account

NewSEO Report now works with real search data — clicks, impressions, position, and keyword analysis

NewTraffic Report now pulls real visitor, session, and traffic source data

NewSite Audit now includes real performance scores and Core Web Vitals metrics

NewAd Spend Analysis can now provide real campaign data and spend reports

ImprovedGoogle Ads Customer ID field added to integration settings

ImprovedSite audit now works without specifying a URL — uses the site address from your integration settings

ImprovedError messages now guide users to the correct settings page

ImprovedAgent now maintains conversation context — short follow-ups like "28 days" or "3 months" re-run the same report with a new period

ImprovedFollow-up suggestions enriched with period options and cross-tool analysis links after reports

ImprovedAgent automatically detects workspace integrations — generates reports without asking for site URL or period

FixedBrowser now only opens when explicitly requested — analytics and report commands no longer trigger browser launch

FixedCache is automatically cleared when integration credentials are updated

FixedFixed integration settings not saving site URL when editing

NewRanking Tracker — track keyword position changes over time with rising, dropping, and new keyword alerts

NewWeb Vitals Alert — monitor page performance scores and get alerts when metrics drop

NewConversion Funnel — analyze your e-commerce purchase funnel with drop-off rates at each stage

NewKeyword Opportunities — discover quick wins, low-hanging fruit, and content gaps from your search data

NewPage Optimizer — identify slow pages with high bounce rates and prioritize fixes by impact

NewAds Keyword Planner — analyze ad keyword performance, find wasteful spend, and discover new opportunities

ImprovedSEO Brief now includes proven keywords from your actual search data for more targeted content planning

ImprovedCompetitor Analysis now shows your own keyword positions alongside competitor marketplace data

ImprovedGEO Report now includes organic search baseline for comparison with AI search visibility

ImprovedE-commerce Analysis now enriched with real conversion and revenue data by traffic channel

ImprovedCampaign Suggestions now based on actual ad performance data for more realistic budget recommendations

21 Mart 2026v4.32

AI Security Hardening

Strengthened AI chat defenses with multilingual protection, smarter input analysis, and extended coverage for all user-generated content.

ImprovedAI chat now detects visual look-alike character attacks across multiple writing systems

ImprovedExpanded multilingual protection — AI chat security now covers more languages and obfuscation techniques

ImprovedTask titles, descriptions, subtasks, and comments are now scanned for suspicious content before saving

ImprovedAI responses are validated against a broader set of safety checks before being shown to users

NewComprehensive automated security test suite covering AI chat input and output validation

21 Mart 2026v4.31

Workspace Intelligence & Agent Reliability

Smarter workspace context for AI, per-capability daily usage limits, improved multi-agent response quality, and enhanced admin monitoring with trend charts.

NewWorkspace Goal — set a description or mission for your workspace so the AI agent understands the bigger picture when making suggestions

NewCapability Daily Limits — set daily usage caps per AI capability with automatic warnings at 80% and hard stops at 100%

NewAdmin Trend Charts — 7-day activity sparklines on each capability card in the admin monitoring dashboard

ImprovedMulti-agent responses are now filtered for quality before being combined, preventing unreliable outputs from affecting the final result

ImprovedMulti-agent operations now include concurrency protection to prevent overlapping executions for the same user

ImprovedWorkspace settings now validate all fields with strict input checks for security and data integrity

ImprovedLanding page now features a cosmic starfield background across all sections for a cohesive space-themed experience

ImprovedOrion constellation planets are now larger and more detailed on desktop displays

21 Mart 2026v4.30

SEO & Operations Suite

Five new AI-powered capabilities for SEO content planning, site auditing, vendor management, ad budget monitoring, and app store tracking.

NewSEO Brief Generator — AI-powered keyword research and content brief creation for blog and category pages

NewSite Audit — comprehensive on-page SEO and content quality analysis for any webpage

NewVendor Follow-up — automated tracking of external vendor tasks with communication history analysis

NewAd Spend Monitor — advertising budget analysis and spend management recommendations

NewApp Store Monitor — app rating tracking with review sentiment analysis and improvement suggestions

21 Mart 2026v4.29

Team Management Intelligence

Three new cross-workspace team management capabilities for workload analysis, workflow optimization, and overdue task tracking.

NewTeam Workload Analysis — cross-workspace team member workload distribution and rebalancing suggestions

NewWorkflow Analysis — kanban column bottleneck detection and process improvement recommendations

NewOverdue Manager — severity-based overdue task reporting with actionable follow-up recommendations

20 Mart 2026v4.28.1

Security Hardening

Comprehensive security review and hardening across the platform. Strengthened input validation, authentication mechanisms, and cross-origin protections.

ImprovedStrengthened form input validation with stricter field length limits and format checks

ImprovedEnhanced admin panel query filtering with strict type and format validation

ImprovedHardened authentication mechanisms against timing-based analysis

ImprovedImproved offline mode to no longer store sensitive session data locally

FixedFixed a redirect validation issue that could direct users to unexpected pages after login

FixedFixed cross-origin message validation for third-party integrations

FixedPrevented outbound integration requests from following redirects to unintended destinations

20 Mart 2026v4.28

Gantt Chart, List View, Custom Fields & Dependencies

A fully interactive Gantt chart for timeline planning, a sortable list view with all task fields, custom fields to track data unique to your workflow, and task dependencies with circular dependency protection. Plus a new start date field for accurate scheduling.

NewGantt chart view — interactive timeline with drag-to-reschedule, zoom levels (day/week/month), dependency arrows, and today marker

NewList view — sortable table with all task fields, custom field columns, and dependency indicators

NewCustom fields — create text, number, select, or date fields to track data unique to your workflow

NewTask dependencies — define blocker/blocked relationships with automatic circular dependency detection

NewTask start date — set both start and due dates for accurate timeline planning

FixedDate display in list and Gantt views now correctly handles all date formats

20 Mart 2026v4.27

Workspace Templates & Multi-Model AI

Ten ready-to-use workspace templates for common e-commerce and project workflows, plus a multi-model AI provider system that lets you connect Claude, OpenAI, or Gemini as your AI backend with automatic failover.

New10 ready-to-use workspace templates — Order Management, Product Launch, Campaign Planning, SEO Project, and more, each with industry-standard columns and labels

NewMulti-model AI provider — connect Claude, OpenAI, or Gemini as your AI backend with automatic failover and circuit breaker protection

ImprovedAdmin panel AI settings — configure provider priority, fallback order, API keys, and cache settings in one place

20 Mart 2026v4.26

Daily Standup Summary & Capability Engineering

A new AI-powered Daily Standup capability that scans the entire workspace and generates a ready-to-use standup report — completed tasks, in-progress work, blockers, and today's priorities. Plus an expanded capability creation standard ensuring every new agent feature ships with full coverage across all system layers.

NewDaily Standup Summary — one-click workspace scan that generates a standup report with completed, in-progress, blocked, and overdue tasks plus today's top 3 priorities

NewOptional focus area filter — narrow the standup to a specific team, topic, or project area

ImprovedCapability creation standard expanded from 14 to 16 mandatory checkpoints — disambiguation labels and category mapping now enforced for every new capability

ImprovedIntent detection now covers standup-related phrases in both Turkish and English with tested negative cases to prevent false triggers

19 Mart 2026v4.25

Agentic Engineering — Spec-First Planning & Code Review

Four capabilities mapped from Karpathy's Agentic Engineering principles: a spec-first task planner that turns vague goals into structured specs and atomic kanban cards; a GitHub PR code reviewer covering security, architecture, and missing tests; a conversation health indicator that warns before context collapse; and a cognitive debt dashboard that tracks failure rates, reruns, and long conversations with a 0–100 efficiency score.

NewPlan & Decompose — describe a goal, get a structured spec with edge cases and risks, then auto-created atomic tasks

NewCode Review — paste a public GitHub PR URL for a security, architecture, and test-coverage analysis

NewContext Health Indicator — amber warning at 60 messages, red danger banner at 80 with a 'Start New Conversation' shortcut

NewCognitive Debt Dashboard — 0–100 score based on failed-run rate, rerun rate, and long-conversation rate; wasted credits tracked

19 Mart 2026v4.23

Memory Insights & Performance Deep-Dive

The Memory tab now shows a live layer breakdown at a glance — click any badge to instantly filter by that layer. Memories with an expiry date display a countdown indicator that turns amber as the date approaches and red when expiry is imminent. The Performance tab gains an efficiency score (credits spent per successful run) on each capability card, plus an expandable top-tools panel that reveals which tools a capability relied on most during the selected period.

NewMemory layer stats bar — live badge counts for each memory layer (identity, preference, workspace, archive) that double as one-click filters

NewMemory expiry countdown — memories with an expiry date now show how many days remain, turning amber within 30 days and red within 7

NewEfficiency score on capability cards — credits consumed per successful run, so you can spot which capabilities give the best return

NewExpandable top-tools panel per capability — click any capability to see the top 10 tools it used, with run counts, success rates, and average duration

ImprovedTool statistics are period-aware — switching the time range always loads fresh data for the selected period, with no risk of stale results carrying over

FixedSeveral reliability improvements to the automation scheduler and performance data consistency

19 Mart 2026v4.22

Automation — Custom Agent Schedules

Agents can now run on a schedule you define. Choose from built-in presets (daily morning, daily evening, every Monday, every Friday, every 3 hours) or enter a custom schedule expression. Active schedules are listed with their next run time and can be deleted at any time. A new notification modal surfaces important agent alerts inline so you never miss a status change.

NewCustom agent scheduling — create recurring runs using preset intervals or a fully custom schedule expression

NewSchedule management UI — view all active schedules with next-run times and remove any schedule with one click

NewIn-app notification modal — important agent status changes surface as inline alerts rather than relying solely on push notifications

ImprovedAutomation tab now shows a real-time schedule list that refreshes automatically when a schedule is created or removed

18 Mart 2026v4.20

Quick Action UX Overhaul & Agent Documentation

Every quick action button now has a dedicated, context-aware input dialog — no more generic prompts. Each button asks exactly the right question before running. Disabled legacy server-side browser tools have been removed from the interface, replaced fully by the Mac-native browser controls. Comprehensive agent capability documentation added covering all 61 capabilities.

ImprovedContext-aware quick action modals — each button now shows its own tailored dialog with a specific title, description, and placeholder instead of a generic prompt

ImprovedSend Notification, Create Product, Product Listing, Update Order Status, XML Feed Sync, Competitor Analysis, and Campaign Suggestion each have their own dedicated input dialogs

ImprovedGenerate Report now runs instantly without requiring user input — workspace context is collected automatically

ImprovedTask Detail, Task History, Create Tasks, and Spotify controls now correctly prompt for input before running

ImprovedRemoved legacy server-side browser automation buttons — the interface now exclusively uses Mac-native browser controls (local browser)

NewComprehensive agent capability reference — all 61 capabilities documented with descriptions, credit costs, timeouts, input requirements, and internal flow

17 Mart 2026v4.19

GEO Analysis — AI Search Visibility

New GEO (Generative Engine Optimization) report to analyze your website's visibility in AI search engines like ChatGPT, Perplexity, Google AI Overviews, and Claude. Discover how to become more visible in AI searches with content structure, citation-worthiness, and technical readiness scores.

NewGEO Report — content optimization and visibility analysis for AI search engines. Content structure, citation opportunities, technical setup, and AI search presence in a single report

NewAI crawler check — automatically verifies your website's accessibility status for GPTBot, ClaudeBot, PerplexityBot, and other AI crawlers

NewAI search presence score — comprehensive scoring system that measures how visible your brand is in AI-generated responses

NewSmart routing — natural language commands like "how do I appear in ChatGPT" or "check my AI search presence" automatically trigger GEO analysis

ImprovedE-commerce quick actions — GEO report integrated alongside SEO and traffic reports for easy access in the same category

17 Mart 2026v4.18

Reliability & Performance Hardening

Comprehensive reliability hardening across the entire agent platform. Scheduled tasks are now more resilient with better resource management and duplicate prevention. Agent performance analytics load faster with smarter data processing. Overall system stability improved with better error recovery and resource cleanup.

ImprovedScheduled task reliability — automated tasks now handle concurrent execution more safely, preventing duplicate runs and ensuring accurate resource tracking

ImprovedPerformance dashboard speed — agent analytics now load significantly faster with optimized data processing, especially for workspaces with extensive history

ImprovedEvening summary intelligence — daily activity summaries are now generated with better context and delivered more reliably

ImprovedProduction deployment safety — enhanced deployment process with automatic health checks and rollback protection

FixedScheduled task recovery — tasks that were interrupted by server restarts now properly clean up and resume on the next cycle

FixedActivity signal accuracy — daily activity tracking no longer creates duplicate entries, ensuring accurate workspace analytics

FixedMulti-agent coordination — the orchestration system now properly resets between consecutive complex requests, preventing stale state from previous runs

FixedTimezone consistency — all daily boundaries for summaries, reports, and cleanup tasks now use consistent UTC-based calculations

15 Mart 2026v4.17

MacBook Smart Actions & Task Quick Actions

Your AI assistant now understands more specific requests for your MacBook and your workspace. New dedicated quick actions for calendar, reminders, notes, contacts, music control, system settings, and email sending. Plus, previously advanced task actions like assigning, moving, and setting priorities are now available as one-tap quick actions.

NewCalendar management — create events, check your schedule, and see today's agenda directly through the agent

NewReminders — create, list, and complete reminders on your Mac with natural language commands

NewNotes — create and search notes in Notes.app through the agent chat

NewContact search — find contact details by name without leaving the conversation

NewMusic control — play, pause, skip tracks, and see what's playing on Spotify

NewSystem control — adjust volume, toggle dark mode, and manage Mac system settings

NewSend email — compose and send emails through Mail.app or Outlook directly from the agent

ImprovedTask quick actions — assign tasks, move to columns, set priorities, due dates, labels, view metrics, and send notifications are now available as one-tap actions

ImprovedSmarter intent detection — the agent now routes your requests more accurately to the right capability, reducing misclassification

15 Mart 2026v4.16.2

Performance & Stability Deep Dive

End-to-end performance optimization and stability hardening across the entire agent system. Smoother real-time streaming, faster conversation loading, and elimination of multiple race conditions that could cause stale data or UI glitches during rapid interactions.

ImprovedSmoother agent streaming — token-by-token message display is now significantly more efficient, eliminating visual stutters in long conversations

ImprovedFaster chat rendering — message bubbles, plan steps, and browser automation cards now skip unnecessary re-renders, making the agent interface noticeably snappier

FixedConversation switching stability — rapidly switching between conversations no longer causes stale messages or plan data to appear from the previous conversation

FixedMulti-agent coordination reliability — sub-agent progress updates now correctly stay within their originating conversation, preventing cross-conversation data leaks

ImprovedConversation loading optimized — loading large conversations is now faster with smarter data retrieval from the server

FixedCancelling a running agent now fully resets the plan view and approval state, preventing ghost UI elements from lingering

ImprovedBetter error feedback — insufficient credits and expired sessions now show clear error messages instead of generic failures

FixedDeleting a conversation while the agent is running now properly stops the active task before removal, preventing orphaned background processes

14 Mart 2026v4.16

Platform Optimization & Infrastructure

Comprehensive platform improvements covering performance, accessibility, security, monitoring, and architecture. API response times reduced, dashboard monitoring expanded, and accessibility standards raised to WCAG 2.1 AA.

ImprovedFaster page loads — smart caching reduces unnecessary API calls by up to 50%, and new database indexes speed up dashboard and list queries significantly

NewDark mode toggle — switch between dark, light, and system-auto themes from the header

ImprovedAccessibility improvements — modal dialogs now properly trap keyboard focus, support Escape to close, and return focus to the triggering element (WCAG 2.1 AA)

ImprovedTask operations feel instant — optimistic updates show changes immediately while syncing in the background

ImprovedEnhanced monitoring — expanded Grafana dashboards with 20+ panels tracking API performance, WebSocket connections, and agent metrics

ImprovedSecurity hardening — sensitive data in logs is now automatically redacted, and security state persists across server restarts

14 Mart 2026v4.15

Response Reliability & PWA Update

Agent responses are now significantly more reliable. Dashboard data extraction is faster with a new direct-read approach. Duplicate message display has been resolved. The PWA update detection system has been improved for quicker notifications.

ImprovedFaster dashboard data extraction — a new direct-read approach retrieves visible metrics without requiring full visual analysis, dramatically reducing wait times

FixedDuplicate message display resolved — a race condition causing user messages to appear twice in certain scenarios has been fixed

ImprovedResponse reliability — added fallback mechanisms ensuring the agent always returns a meaningful response, even when the primary processing path encounters issues

ImprovedPWA update detection — the app now checks for updates more frequently, so you see the update banner sooner after a new version is deployed

13 Mart 2026v4.14

Form Typing Reliability & Dashboard Data Extraction

Browser automation form typing is now significantly more reliable. A critical bug causing every character to be typed twice on certain e-commerce platforms has been resolved. Dashboard metric reading and reporting has been improved. The new conversation button now works with a single click.

FixedCharacter doubling fixed — resolved a technical issue causing each character to be typed twice on certain websites (e.g. Ticimax admin panels)

ImprovedReliable character insertion — form typing now uses Chrome's native Input.insertText method, working consistently across all web frameworks

ImprovedDashboard data reading — the agent now more accurately reads and reports visible metrics (sales, orders, sessions, etc.) from website dashboards

FixedNew conversation button now works with a single click — resolved a timing conflict with the previous conversation still loading

13 Mart 2026v4.13

Stability & Performance

Under-the-hood improvements to browser automation stability and overall performance. Error recovery is more robust and resource usage has been optimized.

ImprovedBrowser automation stability — improved error recovery and session management for more reliable automated browsing

ImprovedPerformance optimization — reduced memory usage and faster processing across multiple agent operations

13 Mart 2026v4.12

Smart Screenshot Optimization

The vision system now intelligently skips unnecessary screenshots, reducing processing time and API costs. Conversation history is compressed to use less memory.

ImprovedSelective screenshot skip — the agent now detects when a screenshot is unnecessary (e.g., after typing) and skips it, saving time and resources

ImprovedHistory compression — older conversation steps are summarized, freeing memory for longer browsing sessions

12 Mart 2026v4.11

Turkish Web Pattern Support

Enhanced support for Turkish websites with locale-aware patterns. Cookie consent dialogs are now handled more reliably across different platforms.

NewTurkish web pattern recognition — the agent now understands Turkish-language navigation patterns, button labels, and form fields on local e-commerce platforms

ImprovedCookie consent handling — improved detection and dismissal of cookie consent banners across a wider range of websites

ImprovedAccessibility tree enrichment — richer element metadata is now available to the AI for more accurate interactions

12 Mart 2026v4.10

E-Commerce Panel Integration

Expanded support for e-commerce administration panels. The agent can now extract structured data from dashboards and return it in organized format. Form filling reliability has been improved.

NewStructured data extraction — the agent can now return organized data (prices, orders, metrics) from web pages in structured format

ImprovedE-commerce panel support — better compatibility with popular Turkish e-commerce platforms and their admin panels

ImprovedForm filling reliability — improved accuracy when interacting with complex form layouts and custom input components

12 Mart 2026v4.9.2

Smarter Visual Marking & Screenshot Accuracy

Visual element marking is now significantly more accurate. Labels no longer overlap in dense navigation areas, modal and dialog content is prioritized, and screenshots now precisely match interaction coordinates on all display types.

ImprovedLabel collision avoidance — numbered element labels no longer overlap in dense navigation bars and button groups

ImprovedScreenshot coordinate precision — screenshots now capture at exact interaction coordinates, eliminating misclicks on high-resolution displays

ImprovedModal awareness — when a dialog or modal is open, only its interactive elements are marked, preventing clicks on background content

FixedNon-clickable elements with disabled pointer interactions are no longer marked, freeing up slots for actual interactive elements

12 Mart 2026v4.9.1

Stability & Reliability Improvements

Under-the-hood improvements to browser automation stability. File operations are now fully asynchronous, temporary files use unique names to prevent conflicts, and several edge cases in page navigation have been resolved.

ImprovedAsynchronous file operations — all file handling is now non-blocking for better responsiveness

ImprovedUnique temporary file paths — eliminates rare conflicts when multiple operations run simultaneously

FixedBack navigation boundary check — going back no longer fails when there is no previous page in history

FixedPage identifier validation — prevents errors from malformed page references during navigation

12 Mart 2026v4.9

Browser Navigation & Memory Optimization

Browser automation is now significantly more capable with back navigation, keyboard shortcuts, and intelligent memory management. The agent can recover from wrong pages by going back, use Ctrl/Cmd shortcuts for form operations, and optimize memory usage during long browsing sessions.

NewBrowser back navigation — the agent can now automatically go back when it lands on the wrong page and try the correct path

NewKeyboard shortcuts — common shortcuts like Ctrl+A, Ctrl+C, Ctrl+V, and Cmd+Z are now supported

ImprovedScreenshots are automatically resized — fast and efficient analysis even on large displays

ImprovedSmart memory management — old screenshots are cleared from memory during long browser sessions

ImprovedSafety time limit — browser automation runs for a maximum of 5 minutes to ensure efficient resource usage

12 Mart 2026v4.8

Visual Marking System & Error Recovery

Browser automation now marks clickable elements on the page with numbered labels for much more accurate clicking. Automatic recovery from temporary errors makes the agent more resilient and reliable.

NewVisual marking system — buttons, links, and form fields are automatically numbered to improve click accuracy

NewAutomatic error recovery — temporary click failures are now automatically retried

ImprovedSmarter element targeting — the agent now targets page elements by number reference, reducing misclicks

ImprovedSecurity errors and temporary errors are handled separately — security blocks stop immediately while transient issues are retried

12 Mart 2026v4.7.1

Offline Experience & PWA Enhancements

Major improvements to offline reliability and Progressive Web App capabilities. Failed actions while offline are now automatically queued and replayed when you reconnect. The app now remembers your active conversation, supports bilingual offline pages, and includes stronger browser security headers.

NewOffline action queue — changes made while offline are automatically saved and synced when your connection returns

NewBilingual offline page — the offline fallback now displays in your preferred language (English or Turkish)

ImprovedAgent chat now remembers your active conversation — navigate away and come back without losing your place

ImprovedOffline and update banners now display in your selected language instead of hardcoded text

ImprovedStronger browser security — added additional security headers to protect against cross-origin attacks

ImprovedExternal scripts now verified with integrity checks before loading

FixedUpdated application version in search engine metadata to reflect the current release

12 Mart 2026v4.7

Infrastructure & Security Hardening

Comprehensive code review and hardening across the entire stack. Security vulnerabilities patched, database queries optimized with intelligent caching, deployment pipeline strengthened with automated health checks and container security. Every API response now includes a version identifier for easier debugging.

FixedStrengthened database query security — all dynamic queries now use parameterized templates to prevent injection attacks

FixedUpdated vulnerable dependencies to their latest secure versions across the entire project

ImprovedFaster task loading — large workspaces now load tasks progressively instead of all at once

ImprovedIntelligent caching — workspace ownership and admin status are now cached, reducing repetitive database lookups by up to 80%

ImprovedUnified connection management — all services now share a single connection pool for better resource efficiency

NewAutomated dependency monitoring — weekly security scans now run automatically and create update proposals

NewPost-deploy health verification — deployments now automatically verify API and frontend availability before completing

NewVersion tracking — the health endpoint now reports the exact deployed version for easier debugging

ImprovedContainer security — all containers now run with restricted privileges and enforced memory limits

11 Mart 2026v4.6.4

Smarter Agent Persistence — No More Early Exits

The browser agent no longer gives up prematurely on challenging pages. When confidence is low, the agent now continues exploring instead of immediately declaring failure. The first few steps are always protected, ensuring the agent tries before deciding.

ImprovedEarly exit prevention — the agent now always tries interacting with the page before deciding a task is impossible

ImprovedExploration-first approach — clicking, scrolling, and hovering actions are no longer blocked by low initial confidence

FixedPremature completion on complex pages — the agent no longer stops at the first screenshot without attempting any action

11 Mart 2026v4.6.3

Full-Screen Browsing & Faster Navigation

The browser now opens in full-screen mode, making menu bars and navigation elements fully visible. Mouse movement is optimized for speed, the correct tab is always selected, and task completion is detected faster when the target page is reached.

NewFull-screen browser — the browser window is maximized automatically so all page elements are visible

NewFast task completion — when the target URL is reached, the task finishes immediately without extra verification delays

ImprovedFaster mouse movement — optimized movement paths reduce click time significantly on nearby elements

ImprovedSmart tab selection — the correct website tab is always chosen, ignoring blank or system tabs

FixedHover loop prevention — the agent no longer gets stuck repeatedly hovering without clicking

11 Mart 2026v4.6.2

Visual Page Change Detection

Browser automation now compares consecutive screenshots to detect whether a page actually changed after an action. This zero-cost visual comparison gives the AI concrete feedback about the effect of each click and scroll.

NewScreenshot comparison — each step now reports whether the page content visually changed after an action

ImprovedAction feedback — the AI receives concrete "page changed" or "page unchanged" feedback to make better decisions

ImprovedChange detection sensitivity — minor visual changes (loading spinners, animations) are distinguished from real page navigation

11 Mart 2026v4.6.1

Improved Menu & Subcategory Navigation

The browser agent now correctly navigates hierarchical menus by hovering parent items before clicking subcategories. This fixes cases where clicking a top-level category led to a wrong page instead of revealing the desired subcategory.

ImprovedHierarchical menu navigation — the agent hovers parent menu items to reveal subcategories before clicking

ImprovedPost-click URL verification — the agent checks that the clicked link led to the expected page

FixedSubcategory navigation accuracy — clicking a top-level category no longer skips the intended subcategory page

11 Mart 2026v4.6

Smarter Waiting & Anti-Detection — Post-Action Intelligence

Browser automation now waits for network activity to settle after clicks and navigation, ensuring pages are fully loaded before the next step. The agent is aware of viewport dimensions for more accurate interactions. Tab management prevents memory issues during long sessions.

NewNetwork-aware waiting — the agent now waits for background requests to complete after clicks and page navigation before taking the next screenshot

NewPage change detection — each action now tracks whether the page content actually changed, giving the AI concrete feedback

NewViewport awareness — the AI now knows the exact browser window size for more precise click and scroll coordinates

ImprovedMore natural browsing rhythm — occasional brief pauses make automated browsing indistinguishable from human behavior

ImprovedTab management — automatic cleanup of inactive tabs prevents memory issues during long browsing sessions

ImprovedBrowser fingerprint diversity — each session uses varied browser signatures to avoid detection patterns

11 Mart 2026v4.5.2

Reliable Tab Tracking & Subcategory Navigation

Browser tabs are now reliably tracked even after navigation. The agent understands hierarchical menus and correctly navigates to subcategories by hovering parent items first. A cancel button lets you stop the agent mid-task.

NewCancel button — stop the agent at any time during browser automation with a single click

ImprovedTab tracking — browser tabs are reliably tracked even after page navigation causes internal ID changes

ImprovedSubcategory navigation — the agent now correctly hovers parent menu items to reveal and click specific subcategories

FixedScroll actions no longer fail when the scroll amount is not specified — sensible defaults are now applied

11 Mart 2026v4.5.1

Reliable Navigation — Verify Before Done

Browser navigation commands are now verified before being marked as complete. The agent takes an extra screenshot after page-changing actions to confirm the page actually navigated, preventing premature task completion. Context-aware intent routing ensures follow-up commands are correctly understood.

NewPost-navigation verification — the agent now confirms page changes actually happened before reporting success

NewContext-aware command routing — follow-up browsing commands like 'go to category' are now correctly understood even without mentioning the browser

ImprovedAction batching safety — if a click causes page navigation mid-batch, remaining actions are safely skipped

FixedBrowser navigation commands no longer report success before the target page has loaded

11 Mart 2026v4.5

Smarter Browsing — Action Batching, Hover, Dropdown & Self-Correction

Browser automation is now significantly faster and smarter. Multiple simple actions can be executed in a single step without intermediate screenshots. Hover interactions reveal hidden menus and tooltips. Dropdown selections are handled natively. The system automatically detects loading states and validates that actions had an effect.

NewAction batching — up to 3 simple actions can now execute in a single step without intermediate screenshots, significantly improving speed

NewHover support — the agent can now hover over elements to reveal dropdown menus, tooltips, and hidden content

NewNative dropdown selection — select/option elements are now handled directly instead of relying on click sequences

NewLoading state detection — the agent now waits for spinners, skeletons, and loading indicators to disappear before proceeding

ImprovedSelf-correction — when actions fail repeatedly, the agent now reflects on what went wrong and tries alternative strategies

ImprovedStep history optimization — long browsing sessions now compress older steps to maintain fast and focused responses

ImprovedPost-action validation — the agent now verifies that clicks and typing actually changed the page before moving on

ImprovedSensitive page detection expanded — more page types are recognized and protected during automation

FixedBrowser tab close commands now work with natural phrases in both Turkish and English

FixedTab close now automatically resolves the active tab when no specific tab is specified

11 Mart 2026v4.4.1

Connection Stability & Screenshot Display

Browser automation connections are now more stable during long sessions with large screenshots. The active page is detected automatically, and browser screenshots are displayed directly in the chat for better visibility.

ImprovedConnection stability — large screenshot transfers no longer cause disconnections during long browsing sessions

ImprovedAutomatic page detection — the correct browser page is identified automatically without manual selection

NewScreenshot display in chat — browser screenshots are now shown directly in the conversation for better task visibility

10 Mart 2026v4.4

Drag & Drop and File Upload Support

Browser automation now supports drag-and-drop interactions and file uploads. Drag actions follow natural Bezier movement paths for realistic behavior. File uploads are protected by multi-layer security validation.

NewDrag-and-drop support — automatically drag and drop elements on web pages

NewFile upload support — automatically select and upload files to file input fields

NewFile upload security layer — file paths are validated across multiple layers to block unsafe paths

ImprovedNatural drag movement — human-like curved movement paths for realistic drag interactions

ImprovedExpanded browser action types — clicking, typing, scrolling, dragging, and file uploading now supported

10 Mart 2026v4.3

Smart Page Detection and Task Chaining

Browser automation now automatically detects interactive elements (buttons, links, forms) on pages for more accurate clicking. When multiple browser tasks run sequentially, session state is automatically transferred between them.

NewSmart page element detection — buttons, links, and form fields are automatically identified for improved click accuracy

NewTask chaining — session data (cookies, login state) is automatically carried over between sequential browser tasks

NewSession transfer — multi-step workflows that maintain login state across browser tasks are now supported

ImprovedClick accuracy improved — page structure coordinates are used alongside screenshot analysis

ImprovedMulti-task reliability improved — previous task results are forwarded as context to the next task

10 Mart 2026v4.2

Human-Like Browser Control — Natural and Intelligent Automation

Browser automation now behaves like a real user. Natural mouse movements, smooth scrolling, realistic typing rhythm, and advanced page-wait strategies deliver much more reliable results.

NewHuman-like mouse movement — realistic clicking behavior with natural curves and timing

NewSmart scrolling — gradual, natural-speed page scrolling with page-by-page navigation for large pages

NewRealistic typing — form filling with word pauses and natural speed variations

NewSpecial key support — Enter, Tab, Escape, arrow keys and more are now supported

NewAutomatic Chrome launch — Chrome can now be started automatically if not running

NewSession persistence — browser sessions can be saved and resumed later

ImprovedScreenshot optimization — smaller size, faster analysis

ImprovedPage load waiting strategy improved — more accurate timing with network traffic monitoring

ImprovedError recovery mechanism added — failed actions are automatically retried

ImprovedSensitive page protection — screenshots skipped on login and payment pages for enhanced security

ImprovedBot detection protection strengthened — browser fingerprint masking and automation traces cleaned

9 Mart 2026v4.1.5

Security & Infrastructure Improvements — Stronger Foundation

Over 60 security and quality improvements applied after a comprehensive security audit. Agent infrastructure modularized and automated test coverage significantly expanded.

ImprovedComprehensive security audit completed — over 60 security and quality improvements applied

ImprovedBrowser automation security layers strengthened — command validation, URL filtering, and content checks tightened

ImprovedDevice command security enhanced — malicious command patterns and character manipulation attempts are now blocked

ImprovedAgent infrastructure modularized — more maintainable and testable architecture

ImprovedAutomated test coverage increased by 75% — 268 automated tests ensuring regression safety

FixedFixed product filtering issue in competitor analysis tool — search results are now correctly filtered

FixedRemoved unnecessary type conversions in e-commerce tools — more consistent and reliable data handling

FixedFixed missing cache eviction that could cause memory leaks

9 Mart 2026v4.1.4

Enhanced Notification & Cookie Management — Seamless Browser Experience

Notification permission prompts and cookie banners shown by websites during browser automation are now dismissed much more reliably.

ImprovedNotification permission prompts are now blocked at the browser level — permission dialogs never appear when websites request notification access

ImprovedLocation, camera, and microphone permission prompts are also automatically blocked — no more intrusive permission dialogs

ImprovedPush notification popup dismissal strengthened — custom popups from popular push notification services are now automatically dismissed

FixedFixed an issue where notification popups could not be dismissed when a cookie banner was closed on the same page

FixedPopups using non-standard button and link elements are now properly detected and dismissed

FixedFixed browser opening requests not being routed to the correct tool — all URL opening commands now work seamlessly

ImprovedRequest routing rules expanded — more commands are correctly recognized and directed to the appropriate tool

8 Mart 2026v4.1

Smart Browser Automation — AI-Powered Web Control

OryoAgent can now control your connected MacBook's browser using AI-powered vision analysis. It takes screenshots, analyzes the page, and performs human-like step-by-step automation.

NewSmart browser automation — human-like web automation on your MacBook browser using AI vision analysis

NewSix new browser tools — page opening, screenshots, click/type/scroll actions, content extraction, and tab closing

NewLive automation tracking — watch each step's screenshot, reasoning, and progress in real time

NewConfigurable connection port — browser connection port can be customized per device

ImprovedMulti-layered security — browser commands pass through three-stage validation at server, adapter, and device layers

ImprovedCredential protection — passwords and sensitive data are masked in AI analysis, logs, and chat history

NewAutomatic cookie and notification management — cookie consent banners and notification permission prompts are automatically dismissed when websites are opened

ImprovedPre-load page protection — smart monitoring system runs before page content loads, catching banners and popups instantly

ImprovedAutomatic JavaScript dialog handling — page alerts, confirmations, and prompts are handled automatically

ImprovedSmart tool routing — browser opening requests are now automatically directed to the correct tool for faster and more reliable automation

ImprovedUnified browser automation — all web operations are securely performed through the user's own browser

8 Mart 2026v4.0

Multi-Agent Orchestration — Specialist Teams for Complex Requests

OryoAgent can now coordinate multiple specialist agents to handle complex, cross-domain requests in parallel.

NewMulti-agent coordination — complex requests are automatically decomposed across specialist agents working in parallel

NewSix specialist agents — task operations, strategic planning, analysis & research, web automation, local device management, and e-commerce

NewIntelligent effort scaling — requests are automatically routed to single-agent or multi-agent mode based on complexity

NewReal-time coordination status — see which specialist agents are working and track their progress live

ImprovedConfidence-based routing — the system assesses decomposition quality and falls back to single-agent mode when uncertain

ImprovedEnhanced observability — detailed metrics and monitoring for multi-agent coordination performance

8 Mart 2026v3.14

AI Context Intelligence — Smarter Agent Interactions

New tools for context visibility, quick command access, smarter resource management, and enhanced safety controls for high-risk operations.

NewContext budget dashboard — see real-time token usage and resource allocation during agent sessions

NewSlash commands — type / in the chat input for quick access to agent capabilities with autocomplete

NewSmarter context loading — agent now loads only relevant tools based on your request, reducing unnecessary processing

ImprovedEnhanced safety controls — high-risk operations now require explicit approval regardless of mode

7 Mart 2026v3.13.5

Infrastructure Hardening — Performance & Monitoring Fixes

Server-side improvements focused on monitoring reliability, security event tracking, and overall system performance. Resolved resource leaks and fixed broken metrics collection.

FixedFixed server metrics collection that had been failing silently, restoring full monitoring and alerting capabilities

FixedFixed agent security event tracking that was silently failing to persist certain events to the database

ImprovedResolved a long-running process leak that was consuming excessive CPU and memory resources on the server

ImprovedServer disk usage optimized — reclaimed significant storage space from accumulated build artifacts

7 Mart 2026v3.13.4

Agent Tool Calling — Root Cause Fix for Empty Parameters

Critical fix for a structural issue that caused OryoAgent to send empty parameters when executing tools like opening apps on MacBook, navigating websites, and running commands. The agent now reliably fills all required parameters.

FixedFixed a fundamental issue where the agent consistently sent empty parameters to tools, causing actions like 'open Safari' or 'navigate to website' to fail repeatedly

ImprovedAgent planning now uses a more flexible parameter format that allows accurate filling of tool-specific values like scripts, URLs, and queries

ImprovedAdded a local device action example to the agent's planning reference, improving first-try accuracy for MacBook automation tasks

7 Mart 2026v3.13.3

Agent Intelligence — Smarter Tool Execution & Error Recovery

Major reliability improvements to OryoAgent's tool execution pipeline. The agent now understands tool parameters better, recovers from errors more gracefully, and avoids getting stuck in repetitive loops.

ImprovedAgent now receives concrete examples when planning tool actions, resulting in significantly more accurate parameter filling

ImprovedWhen a tool action fails due to a minor parameter issue, the agent now attempts an automatic repair before giving up

ImprovedAgent error recovery is now smarter — failed actions include detailed context so the agent can learn from mistakes and create better revised plans

FixedFixed an issue where the agent could falsely detect a 'stuck loop' when retrying the same action with different inputs

ImprovedAll local device tools (file operations, shell commands, email, screenshot) now include usage examples for better agent understanding

7 Mart 2026v3.13.2

Accessibility & Code Quality — Full Compliance Pass

Comprehensive accessibility improvements across all agent components and a complete code quality sweep eliminating all frontend lint warnings.

ImprovedAll animations now respect your system's reduced-motion preference — users who prefer less motion will see static indicators instead

ImprovedDialogs, modals, and interactive elements now have proper screen reader support with descriptive labels and keyboard navigation

ImprovedDelete actions now require a two-step confirmation that auto-resets after 3 seconds, preventing accidental deletions

ImprovedKeyboard users can now see clear focus indicators when navigating through agent conversations and log entries

ImprovedAll hardcoded interface text has been replaced with translatable strings — both English and Turkish are fully covered

FixedResolved all frontend type safety warnings — stronger compile-time checks across the entire codebase

7 Mart 2026v3.13.1

Device Ownership & Screenshot Rendering — Stability & Security Fixes

Devices are now strictly user-owned and accessible from any workspace. Screenshots captured by the agent are rendered as images instead of raw data. Multiple agent execution engine fixes improve reliability.

ImprovedDevices are now personal — your paired MacBook works across all your workspaces, and other workspace members cannot see or access it

NewScreenshots captured by the agent are now displayed as visual images in the chat with click-to-open-fullsize support

FixedFixed an issue where the agent could skip valid actions during retries, causing repeated failures

FixedFixed an issue where temporary failures in one agent run could block all subsequent runs

FixedImproved error messages in agent step cards — specific failure reasons are now shown instead of generic messages

FixedAgent chat conversations remain strictly private — other workspace members cannot access your chat history

6 Mart 2026v3.13

MacBook Local Device Management — OryoAgent Can Now Control Your Mac

OryoAgent can now securely connect to your local MacBook and automate it — run terminal commands, read and write files, control apps via AppleScript, take screenshots, and read emails. Works over a secure reverse-connection so no port forwarding is needed.

NewLocal device connection — pair your MacBook with OryoAgent via a one-time install script. The agent connects securely outbound so it works behind NAT and firewalls

NewShell command execution — the agent can run terminal commands on your Mac and return the output

NewFile system access — read files, write files, and browse directory listings within your approved folders

NewAppleScript automation — control native macOS apps (Finder, Calendar, Notes, etc.) using Apple's built-in scripting language

NewScreen capture — take a screenshot of your Mac's screen and see it directly in the chat

NewEmail reading — read your inbox via Mail.app or Gmail (OAuth2) — subject, sender, date and preview only

NewConnect Mac button — pair your device from the agent header with a status indicator showing online/offline state

ImprovedLocal device capabilities are automatically disabled in the quick actions panel when no device is connected

ImprovedMulti-layer security — command guardrails on both the server and the local daemon prevent destructive operations regardless of agent instruction

6 Mart 2026v3.12.1

Agent Hardening — Human Approval Toggle, Reliability & Security Improvements

Major reliability and security update across the agent system. Introduces a user-controlled supervised mode toggle, persistent failed-operation recovery, and multiple platform-level protections.

NewSupervised mode toggle — users can switch between Autonomous and Supervised mode directly from the agent header. In supervised mode, the agent pauses for human approval before executing plans and risky steps

FixedAgent plan state badge now correctly reflects execution state when stepping through tasks after approval

FixedAgent run failure state now correctly propagates to the plan view badge

NewPersistent failed-operation queue — failed agent operations are now stored durably and survive service restarts for automatic retry

ImprovedCredit system — concurrent run tracking now accounts for in-progress runs to reduce mid-run credit exhaustion risk

ImprovedBrowser agent security — domain allow/block lists can now be configured via environment variables

ImprovedPlugin hook failures now log errors and degrade gracefully instead of silently swallowing exceptions

ImprovedWebSocket sessions now automatically close when the access token expires, preventing stale authenticated connections

ImprovedScheduled agent jobs now use distributed coordination to prevent duplicate execution across multiple server instances

5 Mart 2026v3.11.2

Security Remediation — SQLite Cleanup, Email Hardcoding & Non-Root Nginx

Resolved 3 critical findings from the deep-dive re-scan: deleted local SQLite database artifacts containing PII, replaced hardcoded admin email with environment variable, and fixed Nginx to run on unprivileged port 8080 as non-root user.

FixedSQLite PII risk — deleted local dev.db files and confirmed .gitignore coverage

FixedHardcoded admin email — creditAlertService now reads PLATFORM_ADMIN_EMAIL from environment

FixedNon-root Nginx — switched from port 80 to unprivileged port 8080 with proper path permissions

ImprovedDocker compose port mapping updated to 7779:8080 for frontend container

NewFrontend test infrastructure — 5 test files with 68 tests covering mentionUtils, dateFormatter, tokenStore, inviteTokenStorage, and accessibility

FixedCORS production issue — API container restarted to apply www auto-expansion; www.oryonex.com origin now accepted

5 Mart 2026v3.11.1

Queue Engine Security Audit — 7 Critical Fixes

Deep-dive audit identified and fixed IDOR vulnerabilities, WebSocket data leakage, non-functional queue modes, and missing input validation across the agent queue system. 19 new unit tests added.

FixedIDOR protection — queue endpoints now verify conversation ownership before allowing access

FixedWebSocket privacy — agent conversation events now sent only to the triggering user, not the entire workspace

FixedInterrupt mode — now actually cancels active runs before processing new messages

FixedSteer mode — separated from collect/followup with its own WebSocket event and behavior

FixedQueue drain — items are re-queued instead of silently marked completed when processing fails

FixedLatest drop policy — now correctly drops the incoming message instead of existing queued items

ImprovedQueue config validation — mode, drop policy, capacity, and debounce are now validated at the API layer

New19 unit tests added for queue config validation, drop policy logic, and advisory lock hashing

5 Mart 2026v3.11

OryoAgent — Human Approval Flow, Queue Engine & Model Failover

OryoAgent now requests human approval for critical operations, intelligently queues concurrent messages, and automatically falls back to backup AI models on failure. All decisions are observable via Prometheus metrics.

NewHuman approval flow — plan and step-level approve/reject buttons give users control over critical operations

NewQueue Engine — concurrent messages are safely queued using PostgreSQL advisory locks

New4 queue modes — collect (batch), followup (auto-drain), steer (live injection), interrupt (cancel & restart)

New3 overflow policies — summarize (condense oldest), oldest (FIFO drop), latest (reject new)

NewModel failover — automatic switch to backup AI model when the primary provider fails

New3 new Prometheus metrics — plan approval decisions, failover events, and model usage counters

ImprovedMessage response type refactored to discriminated union — queue and steer states handled separately

ImprovedWebSocket event system expanded — 4 new events (plan approval, tool approval, queued, steered)

5 Mart 2026v3.10.1

OryoAgent — Integration-Powered Analytics & Strategic Analysis (5 New Tools)

5 new AI-powered tools: SEO reports via Search Console, traffic analytics via GA4, XML feed sync, competitor price analysis, and campaign suggestions. Full integration awareness system added.

NewSEO report — Search Console API integration for SEO performance analysis with AI insights

NewTraffic report — GA4 API integration for traffic, sessions, and bounce rate analysis

NewXML feed sync — Google Shopping, Trendyol, and RSS feed parser with diff reports

NewCompetitor analysis — marketplace price search with AI strategic pricing recommendations

NewCampaign suggestion — trend analysis with 3 campaign ideas and implementation steps

ImprovedIntegration awareness system — agent automatically detects active GA4, Search Console, and XML integrations

ImprovedTotal tools: 47 → 49 — all 5 new tools pass 9-point capability checklist

5 Mart 2026v3.10

Security Audit Remediation — Workspace Isolation & Platform Hardening

9 fixes based on comprehensive security audit findings. File upload workspace isolation, server-side session revocation, invitation token masking, Prometheus HTTP metrics, and 16 new security contract tests.

NewWorkspace isolation — file uploads are now tracked via PendingUpload table, preventing cross-workspace attachment binding

NewHTTP metrics — Prometheus histogram and counter for request duration and count tracking with route normalization for low cardinality

New16 security contract tests — covering attachment validation, Zod schema alignment, Prometheus metrics, and PendingUpload model verification

NewAttachment URL validator — shared utility for format, path traversal, and file existence checks

ImprovedServer-side logout revocation — logout now revokes the session on the server before clearing local state

ImprovedInvitation token masking — token field completely removed from API responses

ImprovedAttachment deletion now also removes the physical file from disk — preventing orphan file accumulation

Fixed10 unused catch variables and 3 unused imports fixed — lint warnings reduced from 311 to 296

4 Mart 2026v3.9

OryoAgent Prompt Engineering — Structured Prompt System

All OryoAgent LLM call points have been restructured according to Gemini prompt engineering best practices. Structured XML templates, few-shot examples, self-critique system, and anti-hallucination safeguards deliver more accurate and reliable agent responses.

NewPromptBuilder module — centralized prompt construction for all 4 LLM call points (planning, observation, replanning, synthesis)

NewStructured XML templates — separate sections for role, constraints, context, tools, task, and output format

NewFew-shot examples — successful plan examples for simple queries and multi-step browser workflows

NewSelf-critique system — agent automatically validates tool slug accuracy, param completeness, and dependency ordering before responding

ImprovedAnti-hallucination strengthened — tool results treated as 'absolute limit of truth'

ImprovedConsistent prompt format across all 4 call points — all prompts now use the same XML structure

ImprovedNegative constraints converted to positive patterns — more reliable model behavior

2 Mart 2026v3.8

OryoAgent E-Commerce Tools — Product Management & Sales Analytics

OryoAgent now manages your e-commerce operations. Create products, search your catalog, generate SEO-optimized listings, update orders, and analyze sales — all through natural conversation or quick actions.

NewCreate products — add new items to your workspace catalog with pricing, stock, and category details

NewSearch products — find and filter catalog items by name, category, price range, or stock status

NewGenerate product listings — create SEO-friendly titles, descriptions, and meta tags with AI for multiple platforms

NewUpdate order status — manage order fulfillment with status transitions and tracking information

NewE-commerce analytics — analyze sales trends, inventory levels, and revenue performance with AI-powered insights

ImprovedNew E-Commerce category tab — dedicated section for all commerce-related capabilities

ImprovedProduct and order data models integrated into agent workspace context

2 Mart 2026v3.7

OryoAgent Task Editing — Comments, Subtask Management, and More

OryoAgent can now fully edit your tasks without spending any credits. Add comments, update titles and descriptions, create subtasks, toggle completion status, and manage labels — all through natural conversation.

NewAdd comments — ask the agent to add notes, remarks, or annotations to any task

NewEdit tasks — change task titles, descriptions, or meeting links through conversation

NewCreate subtasks — add individual subtasks to tasks by simply telling the agent

NewToggle subtasks — mark subtasks as done or reopen them with a simple command

NewRemove labels — detach labels from tasks when they're no longer needed

ImprovedAll 5 new actions are completely free — zero credits consumed

ImprovedReal-time sync — all changes are instantly reflected across all connected devices

1 Mart 2026v3.6.1

Production Polish & Deployment Improvements

Final pre-deployment review and cleanup pass. Improved logging consistency across the agent system and removed unused configuration from the build pipeline.

ImprovedAgent system logging standardized — all internal warnings now use the structured logging system for better observability

ImprovedBuild pipeline simplified — removed unused configuration entries from the frontend build process

ImprovedComprehensive pre-deployment review completed — all systems verified across backend, frontend, and security layers

1 Mart 2026v3.6

Performance Testing & Production Hardening

Comprehensive load testing infrastructure ensures OryoAgent performs reliably under real-world conditions. Detailed monitoring dashboards provide full visibility into system health during peak usage.

NewLoad testing suite — automated performance tests covering single runs, concurrent operations, stress scenarios, and recovery validation

NewPerformance dashboard — real-time monitoring of agent latency percentiles, error rates, and resource utilization during load tests

ImprovedSystem resilience — validated circuit breaker behavior, rate limit enforcement, and graceful degradation under heavy load

ImprovedRecovery monitoring — automatic verification that the system returns to baseline performance after traffic spikes

1 Mart 2026v3.5

OryoAgent Advanced Intelligence — Smart Reports, Workflow Insights & Resilience

OryoAgent gains powerful new analytical capabilities with four advanced AI tools, plus enhanced reliability with automatic failover, hybrid search, and a plugin system for extensibility.

NewReport generation — create comprehensive sprint, weekly, or monthly reports with AI-powered analysis of team progress and metrics

NewWorkflow suggestions — get smart recommendations for process improvements, bottleneck resolution, and automation opportunities

NewBlocker detection — automatically identify impediments, stuck tasks, and blockers across your workspace with severity scoring

NewAuto triage — AI-powered task prioritization that analyzes urgency, effort, and impact to suggest optimal priority assignments

ImprovedSearch quality — enhanced memory search combining multiple strategies for more accurate and relevant results

ImprovedAI reliability — automatic failover between AI providers ensures uninterrupted service during outages

ImprovedExtensibility — new plugin system allows workspace-level customization of agent behavior

ImprovedError recovery — failed operations are automatically queued for retry with intelligent backoff

1 Mart 2026v3.4

OryoAgent Proactive Automation — Scheduled Tasks and Smart Alerts

OryoAgent now works for you even when you're not around. Automated schedules monitor your workspace, detect overdue or stale tasks, create daily digests, and provide sprint health insights — all proactively.

NewProactive monitoring — automated checks run throughout the day to detect deadlines, overdue tasks, and stale work items

NewDaily digests — morning summaries with priorities, completed items, and blockers are created automatically

NewSprint health analysis — periodic progress tracking with velocity estimates and risk scoring

NewWeekly summaries — all daily activity is consolidated into weekly overviews for long-term tracking

NewAutomation settings — enable, disable, or manually trigger each schedule from workspace settings

NewSmart notifications — team members are automatically notified about approaching deadlines and overdue tasks

ImprovedConversation handling — long conversations are now automatically summarized to maintain quality responses

ImprovedResponse feedback — a helpful indicator now appears when the agent takes longer than usual to respond

ImprovedCredit management — daily credit limits for automated tasks with automatic reset and smart exemptions

1 Mart 2026v3.3

OryoAgent Persistent Memory and Daily Digest System

OryoAgent now remembers what it learns across conversations. Your preferences, work patterns, and project knowledge are securely stored to provide personalized, context-aware responses in every new conversation.

NewPersistent memory — OryoAgent remembers insights across conversations for a seamless experience

NewLayered knowledge — personal preferences, project details, and workspace data are managed at different priority levels

NewSmart search — fast, relevant text-based search across all stored knowledge

NewDaily digest — record priorities, completed work, notes, and blockers for each day

NewMemory panel — a new interface to view, search, and manage stored knowledge

ImprovedContext-aware responses — OryoAgent automatically uses past learnings in new conversations

ImprovedAuto-cleanup — old and unused knowledge is automatically archived after set periods

ImprovedPrivacy — memory is fully workspace-scoped, isolated from other workspaces, and deletable anytime

1 Mart 2026v3.2

OryoAgent Direct Actions — 8 New Capabilities Without Credits

OryoAgent can now directly assign tasks, move them between columns, change priorities, set deadlines, attach labels, search tasks, view workspace metrics, and send notifications — all without using any credits.

NewAssign task — assign or unassign a team member to any task

NewMove task — move tasks between columns (e.g. To Do to In Progress)

NewSet priority — change task priority to High, Medium or Low

NewSet due date — set, change or clear task deadlines

NewAdd label — attach existing workspace labels to tasks

NewSearch tasks — find tasks by text, assignee, priority, column or overdue status

NewWorkspace metrics — view task statistics, completion rates and team workloads

NewSend notification — notify a team member about task changes or reminders

ImprovedAll 8 new actions are zero-cost — no credits consumed

ImprovedReal-time board sync — all changes instantly reflected across connected clients

1 Mart 2026v3.1

OryoAgent Streaming Responses and Smarter Resource Usage

OryoAgent now streams responses in real time as they are generated. Behind the scenes, intelligent resource management ensures faster and more efficient operations.

NewStreaming responses — see the agent's reply appear word by word in real time

NewLive typing indicator with pulsing cursor during response generation

ImprovedSignificantly reduced resource usage per operation through smarter internal processing

ImprovedTool instructions loaded progressively — only relevant details are included at each stage

ImprovedOlder step results are automatically compacted to keep conversations fast and focused

ImprovedAll optimizations are backward compatible — existing workflows are unaffected

1 Mart 2026v3.0.1

OryoAgent Recipe System — Save and Rerun Repeated Workflows

OryoAgent now lets you save frequently used workflow steps as recipes. Saved recipes can be rerun with a single click, saving both credits and time.

NewRecipe creation — save workflow steps with titles, descriptions and trigger phrases

NewRecipe list — view, edit and manage your workspace or personal recipes

NewVisibility control — keep recipes private or share them with your entire workspace

NewRecipe matching — previously saved recipes are automatically suggested for similar workflows

NewReal-time notifications — instant updates when a recipe is used or saved

ImprovedPersonal recipes are only visible to their owner — no information leaks to other users

ImprovedEmpty or whitespace-only values in recipe steps are now rejected

ImprovedAllowed workflow types are validated to reject invalid input

1 Mart 2026v3.0

OryoAgent v3 — Multi-Step Intelligent Execution Loop

OryoAgent now handles complex tasks step by step — planning, executing and verifying each action. The new architecture delivers more reliable results with detailed progress tracking.

NewMulti-step execution loop — the agent plans, executes and verifies complex tasks step by step

NewTask plan view — follow the agent's planned steps in real-time as they execute

NewState machine — every execution state transition is guarded by safety checks

NewStep-level progress cards — view the status, duration and result of each step individually

NewNew architecture visualization added to the admin monitoring panel

ImprovedExisting workflow fully preserved — the new system is backward compatible

ImprovedTimeout and cancellation support — long-running operations are safely terminated

ImprovedCredit check before each step — budget overruns are prevented

28 Şubat 2026v2.24

OryoAgent Smart Conversations — Suggestions, Feedback and Deep Links

OryoAgent now suggests your next steps after every response, lets you jump directly to created tasks, and allows you to provide feedback on response quality.

NewSmart follow-up action suggestions — start your next step with a single click after each agent response

NewExpandable reasoning panel — see how the agent thought about its response

NewOne-click access to agent-created tasks — jump directly from chat to the task

NewThumbs up/down feedback on agent responses

NewSafety layer preventing the agent from performing destructive operations — created tasks, comments and other items cannot be accidentally deleted

ImprovedFollow-up action suggestions now automatically adapt to the user's language

ImprovedImproved conversation context quality — more accurate responses in longer conversations

FixedFixed feedback selection being lost when switching feedback on error

FixedFixed feedback buttons appearing on messages not yet saved to server

FixedFixed page scrolling to top when clicking the already active sidebar tab

FixedEliminated page jumping to top when switching between sidebar tabs

27 Şubat 2026v2.23

OryoAgent Natural Chat — Workspace Task Awareness

OryoAgent can now see your workspace tasks. Ask questions in natural language about column-based task lists, priorities, assignments and overdue items.

NewWorkspace task awareness — OryoAgent can now see all tasks across columns with priorities, assignments and due dates

NewNatural language chat — ask questions like 'What tasks are in To Do?', 'Which is the most important task?', 'Any overdue items?' and get answers based on real data

NewColumn-based task list — up to 15 tasks per column with title, priority, assignee and due date are provided to the agent

ImprovedTask-linked conversations now include both task details and workspace overview — agent sees the full context

27 Şubat 2026v2.22

Comprehensive Bug Audit and Security Fixes

Comprehensive code scan performed. Security vulnerabilities, performance issues, API mismatches and Service Worker errors fixed.

FixedSecurity improvement on password reset page

FixedAtomic safety added to credit transactions — double-spend prevented

FixedInvitation acceptance consistency strengthened — concurrent duplicate memberships prevented

FixedChat message API compatibility fix applied

FixedOryoAgent conversation creation lock added — prevents duplicate conversation creation on rapid double-send

ImprovedSubtask update performance improved — bulk update support added

ImprovedSafety limits added to admin queries

FixedService Worker error handling improved — unhandled errors in caching operations fixed

FixedService Worker cache issue fixed — unnecessary cache cleanup eliminated

ImprovedMessage cleanup service switched to async — file deletion operations no longer block the event loop

FixedReact key warnings fixed across multiple components — unique keys now used

27 Şubat 2026v2.21

OryoAgent Super Admin Monitoring Dashboard

Agent monitoring, architecture visualization and security event panel added for platform admins.

NewOryoAgent monitoring tab added to Admin Dashboard — visible only to platform admins

NewOverview panel — total runs, success rate, credits used, conversations, security events and time series chart

NewCapability detail cards — run count, success rate, avg execution time and artifact info for all 6 capabilities

NewSecurity events table — injection attempts, blocked messages, suspicious outputs persisted to DB and displayed in admin panel

NewArchitecture flow diagram — visual representation of the agent system workflow

NewUsage rankings — top users and workspaces by agent usage

NewSecurity events now persistently stored in database

NewNew admin API endpoints added for agent monitoring

27 Şubat 2026v2.20

OryoAgent UX Improvements and Quick Action Fixes

Markdown rendering for agent messages, metadata badge fix, quick action UX improvements, and capability context filtering added.

NewMarkdown rendering for agent messages — headers, lists, bold/italic text, code blocks, and inline code support added

FixedAgent message metadata badges fixed — created item badges now display correctly

ImprovedQuick action messages improved — descriptive and meaningful instruction text added for each capability (e.g., 'Analyze this task thoroughly...')

NewCapability context filtering — task view shows only task-context capabilities (Analyze, Subtasks, Progress), workspace shows all capabilities

FixedQuick action results crash fix — missing data fields added to prevent errors

FixedTask ordering sync across clients fixed after task creation

26 Şubat 2026v2.19

OryoAgent Security v5 - Deep Dive Scan Fixes

Multiple findings fixed based on comprehensive security scan. Unicode defense, output sanitization, access control fix, and AI safety settings added.

FixedOutput security checks added to conversational responses — all AI outputs are now automatically validated

NewAdvanced character normalization added — lookalike character bypass attempts now blocked

ImprovedCharacter validation scope expanded — more bypass attempts now detected

ImprovedNested malicious input cleaning strengthened

ImprovedInput sanitization layer strengthened — additional bypass prevention measures added

FixedUser data secured in prompt context — injection via database content prevented

ImprovedConversation history isolation strengthened

FixedAgent run data access control strengthened — users can now only see their own data

NewAI API safety filters activated

ImprovedUnnecessary logs hidden in production environment

ImprovedAdditional security checks added to markdown output

ImprovedAI error messages sanitized — internal details no longer exposed to users

26 Şubat 2026v2.18

OryoAgent Error Masking & Log Security

Agent endpoint error messages sanitized, AI error details hidden from users, and sensitive information redacted in security logs.

ImprovedError messages sanitized across agent endpoints — internal details no longer exposed to users

ImprovedAI API errors masked — users see generic messages

ImprovedSensitive information masked in security logs

26 Şubat 2026v2.17

OryoAgent Security v4 - Full OWASP Compliance

Comprehensive security update based on OWASP standards. Text manipulation, output protection, and sensitive information leak defenses added.

NewText manipulation attack detection and blocking added

NewSensitive information leak detection added to AI output

NewPotential data exfiltration vectors in output content are now sanitized

ImprovedInput normalization strengthened — additional bypass prevention layers added

ImprovedComprehensive security checklist created — based on OWASP standards

26 Şubat 2026v2.16

OryoAgent Advanced Security v3

Comprehensive security update based on OWASP standards. Advanced attack detection, suspicious activity tracking, and AI output limits added.

NewReversed text bypass detection added

NewConversation-level suspicious activity tracking added

NewAI output leakage pattern detection added

ImprovedAI output generation limits added — unbounded generation prevented

ImprovedSystem prompt hardened — additional attack vector protections added

ImprovedData size sent to AI is now bounded

ImprovedError messages sanitized — users see generic messages

26 Şubat 2026v2.15

OryoAgent Security Hardening

Comprehensive security layer added for OryoAgent. Prompt injection detection, system instruction protection, output sanitization, and audit logging make your AI assistant more secure than ever.

NewPrompt injection detection engine — automatically blocks attacks across multiple languages

NewDatabase content sanitized before sending to AI

NewConversation creation rate limiting added

NewSecurity audit logging — suspicious attempts are automatically logged

NewIndirect injection detection — social engineering patterns automatically caught

ImprovedSystem instructions hardened — user data securely isolated

ImprovedSystem prompt leak detection active across all AI capabilities

ImprovedConversation history passes through security checks before sending to AI

26 Şubat 2026v2.14

OryoAgent - AI Assistant System

Interact with OryoAgent using natural language. Perform task analysis, subtask generation, risk scanning, and sprint planning powered by AI.

NewOryoAgent panel in the sidebar — full-page two-panel interface for chat and run log tracking

New6 AI capabilities: Analyze, Generate Subtasks, Progress Report, Create Tasks, Risk Scan, Sprint Plan

NewNatural language chat — manage your tasks by conversing with the agent

NewRun logs — detailed tracking of credit usage, duration, status, and created artifacts

NewConversation management — search, filter, and delete past conversations

NewQuick actions — launch workspace-wide AI analysis with a single click

NewOryoAgent section in task modal — interact with the agent directly from any task

ImprovedReal-time agent status updates via WebSocket

ImprovedMobile-responsive design with tab-based sidebar and chat switching

26 Şubat 2026v2.13

Security, SEO & Task Modal Redesign

Production console security hardened, SEO infrastructure improved, and Attachments, Subtasks, and Comments sections in the task edit modal modernized.

ImprovedAttachments, Subtasks, and Comments sections redesigned with card layout

NewProgress bar added for subtask completion tracking

ImprovedSection headers now include icons and count badges

ImprovedSubtask input redesigned as a combined input+button group

ImprovedEmpty state messages added when no attachments or subtasks exist

FixedProduction environment security improvements applied

ImprovedEmail addresses masked in server logs for enhanced security

ImprovedCanonical URL, hreflang, and meta tag management improved

26 Şubat 2026v2.12.1

Drag & Drop Performance Fix

Fixed lag and unwanted scrolling issues when dragging tasks on the Kanban board.

FixedFixed lag and jank when dragging tasks on the board

FixedFixed unwanted upward scrolling during task drag

24 Şubat 2026v2.12

In-App Notification Center

Stay on top of your workspace activity with the new notification center. See comments, task updates, and more — grouped by date, clickable to open the related task instantly.

NewNotification bell icon in the header with real-time unread badge

NewActivity feed showing comments, task creations, updates, moves, and subtask additions

NewNotifications grouped by date — Today, Yesterday, and older dates

NewClick any notification to open the related task directly

ImprovedHover effects and chevron indicator for clickable notifications

ImprovedOwn actions are filtered out — you only see activity from teammates

24 Şubat 2026v2.11

Drag & Drop File Upload

Upload files to task attachments by simply dragging and dropping. Multi-file support and instant visual feedback make file uploads much faster.

NewDrag & drop file upload support in task modal

NewMultiple files can be uploaded at once

NewDrop zone highlights with visual feedback on drag over

ImprovedAttachment upload area redesigned for better usability

ImprovedFile size and format info shown directly in upload area

24 Şubat 2026v2.10

PWA & Settings Panel Improvements

All members can now access the Settings panel. Customize theme color, font, and display preferences. PWA infrastructure strengthened.

NewSettings panel now accessible to all workspace members (Labels, Calendar, Display)

NewTheme color customization — choose from 8 presets or enter any hex code to personalize buttons, badges, and accents

NewFont selection — Inter, Roboto, Poppins, Nunito, JetBrains Mono and more

NewFont size setting — Small, Normal or Large

ImprovedUser management, invitations and workspace reset restricted to admin role only

NewPWA: Offline support, background sync and smart caching strategies

NewPWA: Notification actions — View or Dismiss buttons

ImprovedSecurity: CSP headers, upload security and notification origin validation

FixedFixed duplicate notification issue on iPhone

FixedFixed service worker update race condition

22 Şubat 2026v2.9

@Mention User Tagging

Tag workspace members in task comments using @ and send them instant notifications.

NewTag workspace members by typing @ in comments

NewAutomatic notifications sent to mentioned users

ImprovedComments save instantly without needing to click Save

ImprovedMentions displayed as clean @Name format in textarea

NewQuick member selection with keyboard (arrow keys + Enter)

22 Şubat 2026v2.8

Server-Side Notification Tracking

Unseen task notifications on the Kanban board are now tracked server-side. Cross-device sync and persistent state support.

NewUnseen task tracking moved to server-side

ImprovedNotification state now persists after hard refresh

ImprovedReal-time notification updates via WebSocket

FixedFixed localStorage-based notification corruption issues

ImprovedTask description field is now vertically resizable

21 Şubat 2026v2.7

SEO & Performance Improvements

Major improvements in search engine optimization and page load performance.

FixedRemoved duplicate hreflang tags

NewAdded trailing slash redirect

ImprovedFont loading optimization to prevent render blocking

FixedResolved Google Search Console indexing issues

FixedImproved Bing Webmaster Tools compatibility

20 Şubat 2026v2.6

Email Verification & Admin Panel

Email verification system for account security and admin panel improvements.

NewEmail verification system added

ImprovedAdmin panel UI updated

ImprovedMulti-language support expanded

15 Şubat 2026v2.5

Chat Encryption Indicator

Added encryption indicator so users can see their messages are secure.

NewEncryption indicator added to chat header

ImprovedMessage security visually communicated to users

10 Şubat 2026v2.4

Score Visibility & Contact Form

User score visibility settings and new contact form.

NewScore visibility settings (public/private)

NewContact form added

FixedSeparated game scores from work scores

1 Şubat 2026v2.3

Asteroid Dodge Game Module

New Asteroid Dodge game added to the gamification system. Weekly leaderboard and mobile support.

NewAsteroid Dodge game module

NewWeekly leaderboard

ImprovedFullscreen mode and mobile responsive UI

ImprovedDirect movement joystick control

20 Ocak 2026v2.2

Beta Period & FAQ Update

Transition to beta period with pricing and FAQ updates.

NewBeta period FAQ added

ImprovedPricing page updated to beta/coming soon

ImprovedButton links and content updates

10 Ocak 2026v2.1

Notification System & Mobile Menu

Push notification infrastructure and mobile hamburger menu added.

NewPush notification system added

NewMobile hamburger menu

ImprovedMulti-language support improvements

20 Aralık 2025v2.0

Real-Time Chat System

Real-time chat system with encrypted messaging added.

NewEncrypted real-time chat system

NewNew marketing pages (Security, FAQ, About, Contact)

ImprovedSEO and geo-targeting improvements

ImprovedMobile responsiveness enhancements

21 Aralık 2025v1.2.1

Account Deletion — Workspace Handling

Account deletion now properly handles workspace ownership. Single-member workspaces are deleted, multi-member workspaces transfer ownership to the next admin.

NewAutomatic workspace handling during account deletion — single-member workspaces deleted, multi-member workspaces transfer ownership

FixedAccount deletion now works for users who own workspaces (previously blocked with error)

8 Aralık 2024v1.2

Internationalization & SEO Improvements

Full i18n support for Turkish and English, comprehensive FAQ sections, AI Authority Checklist, and structured data for SEO.

NewFull internationalization (i18n) for Turkish and English

NewComprehensive FAQ sections across all marketing pages

NewAI Authority Checklist for SEO and Knowledge Graph optimization

NewSchema.org structured data and hreflang tags for multi-language SEO

ImprovedUpdated pricing model and improved mobile responsiveness across all components

FixedHardcoded text replaced with i18n across LandingPage and ContactPage

24 Kasım 2024v1.1

Mobile Responsiveness & SEO

SEO/GEO improvements for marketing pages, mobile responsiveness for admin components, and notification badge improvements.

NewSEO/GEO improvements for marketing pages

NewYellow badge indicator for unseen task updates

ImprovedMobile responsiveness for admin components and navigation links

FixedNavigation link issues and badge visibility logic fixed

1 Aralık 2025v1.0

Initial Release - AI-Powered Kanban

First release of OryoNex: AI-powered Kanban project management.

NewAI-powered task management and brainstorming

NewDrag-and-drop Kanban board

NewWorkspace and team management

NewCredit-based AI system

NewTheme support (light/dark)