v2.0.1 — Security Filter Confirmation & Version Sync Fix

🛡️ v2.0.1 — Security Filter Confirmation Popup & Version Sync Fix

What's New

🔔 Security Filter: User Confirmation Instead of Hard Block

Previously, Yengi's built-in security filters would silently block terminal commands
if the project type didn't match (e.g. running python in a non-Python folder).

Now, instead of a hard block, Yengi shows a confirmation popup explaining exactly
what was detected — and lets you decide whether to proceed.

Affected commands:

• python / pip — checks for .py files, requirements.txt, setup.py, pyproject.toml
• npm / yarn — checks for package.json
• flutter — checks for pubspec.yaml
• dotnet — checks for .csproj / .sln

🔧 Version Sync Fix

The About window was showing v1.16 even after upgrading to v2.0.0.
UpdateService.CurrentVersion was hardcoded and out of sync with the actual build version.
Fixed: All version references now correctly report v2.0.1.

---

📦 Installation

Download Yengi_Setup_v2.0.1.exe below and run the installer.
The built-in auto-updater will also notify existing users automatically.

v2.0.0 — Autonomous Browser Agent & Yengi Router v2.0

🚀 Yengi v2.0.0 — Autonomous Browser Agent & Mobile Store Deployment Engine

We are thrilled to announce Yengi v2.0.0, our biggest milestone update to date! This release transforms Yengi into a truly agentic IDE with an embedded Autonomous Browser Agent running directly inside your workspace, an automated Mobile & Store Deployment Engine, and the newly fine-tuned Yengi Router v2.0 covering all 39 native IDE tools.

---

🌟 Key Highlights

🌐 1. Embedded Autonomous Browser Agent

Yengi can now directly interact with the web without launching distracting external pop-up windows:

• Integrated Editor Tab: Browser sessions now render natively as a dedicated 🌐 Tarayıcı tab inside the editor panel, alongside your code files.
• Chromium Engine (WebView2): High-performance rendering with desktop User-Agent emulation and persistent cookies/sessions via Yengi/BrowserAgentCache (login once to Google Play Console, GitHub, or App Store, and the AI agent retains access).
• Zero-Deadlock Lifecycle: Completely redesigned WebView2 initialization using a modern Loaded event pattern to guarantee zero UI freezes.

📱 2. Mobile & App Store Deployment Engine (New!)

Automate the tedious process of publishing and updating mobile apps on Google Play Console and Apple App Store:

• PrepareStoreAssets (New Tool): Automatically transforms raw logos, banners, and game screenshots into pixel-perfect store specifications:
  • Google Play Store: 512x512 app icons, 1024x500 feature graphics, and 1080x1920 / 16:9 screenshots.
  • Apple App Store: 1024x1024 app icons, 1290x2796 iPhone display screenshots, and 2048x2732 iPad displays.
  • Generates assets directly inside your active project's store_assets/ directory.
• BrowserUploadFile (New Tool): Injects local binary packages (.aab, .apk, images, videos, zip files) directly into HTML5 web forms (DataTransfer & File API) without manual OS file-picker dialogues.
• Human-in-the-Loop Checkpoints: Pauses gracefully for 2FA / login verification, then resumes automated form filling, store listing updates, and pre-submission visual reviews via TakeScreenshot.

🧠 3. Yengi Router v2.0 (Fine-Tuned Local 1.5B Model)

• 100% Native Tool Coverage (39 Tools): Re-trained from scratch on 2,400+ bilingual scenarios covering planning, checkpoints, browser agent, tasks, store deployment, and memory.
• Ultra-Fast & Accurate Decision Making: Zero-fallback, pinpoint tool selection in ~0.2s with ~1.2 GB RAM footprint.
• Direct Trust Architecture: Removed forced tool-injection fallbacks; the IDE now trusts Yengi Router's decisions natively.
• Available now on Hugging Face: mdaiworks/yengi-router-1.5b (Q4_K_M ~940 MB).

---

🛠️ Complete Browser & Deployment Tool Suite

• BrowserNavigate: Autonomous navigation to any URL with load verification.
• BrowserGetDOM: Reads and parses web page DOM structure, text contents, links, and forms.
• BrowserClick: Clicks elements by CSS selector or human-readable text (e.g., "Sign In", "Submit").
• BrowserFill: Fills input boxes, text areas, and submits forms.
• BrowserScroll: Smooth page scrolling to discover lazy-loaded content.
• TakeScreenshot: Captures high-res snapshots of web pages for visual multimodal evaluation.
• BrowserUploadFile: Programmatic local file uploading to web forms and portals.
• PrepareStoreAssets: Automated pixel-perfect scaling for mobile store assets.

---

🐛 Bug Fixes & Stability

• Fixed WPF WebView2 initialization deadlock on background task threads.
• Prevented unintended tool-clearing in conversational casual chat filters when web intents are detected.
• Cleaned up obsolete window wrappers in favor of lightweight UserControl tab components.
• Output paths for generated store assets now strictly bind to the active open project workspace.

---

📦 Download & Installation

• Download the standalone installer Yengi_Setup_v2.0.0.exe below.
• Self-contained .NET 10 desktop runtime included (no extra framework installation required).

🚀 Yengi v1.16 — Dev Release

🚀 Yengi v1.16 — Dev Release: ControlNet 2D Game Sprite Engine & Image Studio Upgrades

We are thrilled to announce Yengi v1.16 (Dev Release)! This release brings a massive leap forward for game developers, technical artists, and prompt engineers. Introducing the ControlNet 2D Game Sprite Engine, Deterministic Image Editing, Seed Locking for Character Consistency, and Full 3-Language (English, Turkish, Chinese) UI & Terminal Localization.

---

✨ What's New in v1.16

🎨 Image Generation Studio Upgrades

• 🌱 Seed Locking for Character Consistency: Easily lock your seed value (🌱) to maintain visual continuity, style, and character design across multiple prompts.
• ✏️ Deterministic Live Interactive Editing: Toggle live edit mode with the glowing pencil button (✏️). Yengi intelligently preserves your original subject prompt while applying multi-turn refinements, eliminating prompt conflicts and inflation.
• Compact Action Bar: Quick action buttons (🌱 Seed, ✏️ Edit, 📁 Show in Folder, 📋 Copy Bitmap, 🔄 Regenerate) right below generated images.
• Live Pulse Progress Indicator: Smooth animated feedback (🎨 Rendering image, please wait...) so you always know when local open-weights models (or cloud APIs) are actively rendering.

🕹️ ControlNet & 2D Game Sprite Engine

• ComfyUI & Cloud ControlNet Dispatcher: Seamlessly connect local ComfyUI instances (http://127.0.0.1:8188) or fallback cloud APIs for pose-guided character generation.
• Automatic Pose Keyword Recognition: Automatically detects character motion keywords (walk, run, attack, jump, idle, t-pose, sprite-sheet) and applies OpenPose control maps.
• Workspace Settings Control: Configure provider URLs, default pose maps, and ControlNet toggles directly in Workspace Settings (⚙️).

🌐 Full 3-Language Localization (EN / TR / ZH)

• Complete localization coverage across all UI Buttons, Hover Tooltips, Terminal Execution Logs, and Status Badges for English (EN), Turkish (TR), and Chinese (ZH).

---

🛠️ Technical Improvements

• Optimized prompt history parser to prevent instruction bleed over multi-turn edits.
• Fully updated WPF components compiled cleanly on .NET 10 (LTS) with 200+ passing unit tests.
• Packaged single-file setup installer Yengi_Setup_v1.16.exe (Self-Contained x64).

---

📦 Installation & Quick Start

1. Download Yengi_Setup_v1.16.exe below.
2. Run the installer and open Yengi.
3. Enjoy the next-generation AI IDE experience!

---

Thank you for supporting Yengi! If you like this update, please consider giving us a ⭐ Star on GitHub!

v1.15 — Local Visual Model Settings

🚀 New Features

Local Visual Model Settings (Resolution & Step): Added Resolution options (512x512, 768x768, 1024x1024, etc.) and Step count (15, 20, 30, etc.) to the Visual Studio Settings (⚙️) menu.

Mac / Apple Silicon MLX Qwen-Image Support: Local Qwen-Image / Flux models now reduce generation time from 10–15 minutes to ~1.4 minutes (~87 seconds) with 512x512 resolution and 15–20 steps.

Modern Dark ComboBox Design: Updated the default Windows XP white dropdown theme to a VS Code–compatible dark theme (#1e1e1e background, #007acc hover, #00ffb7 selected).

🐛 Bug Fixes

WPF ComboBoxItem Text Parsing: Removed object prefix (System.Windows.Controls.ComboBoxItem) to ensure selected resolution and step values are correctly passed to the local server (rapid-mlx / mflux).

HttpClient Timeout Extended: Increased HTTP request timeout to 10 minutes to prevent errors during local MLX image generation.

Pollinations URL Resolution Parameter: Selected resolution is now correctly applied to URL parameters (width & height) when using the free gateway.

v1.14 — Bug fix

🚀 Process Execution & Dev Server Keep-Alive (v1.14)

⚡ Non-Blocking Terminal Output Streaming

• Upgraded process standard output reading to line-by-line event streaming (OutputDataReceived).
• Fully eliminated terminal hangs and 5-minute blocking delays when running long-running dev servers.

🛡️ Dev Server Keep-Alive

• Commands like npm run dev, npm run zipper:dev, vite, or http-server now remain active on local ports in the background.
• Yengi automatically captures dev server startup logs within 15 seconds, marks the status as [BACKGROUND DEV SERVER ACTIVE], and seamlessly allows the AI agent to continue its workflow without halting.

---

Built with .NET 10 (LTS) | WPF | 201/201 Passing Unit Tests

v1.13 — Smart Dev Server Timeout & Background Process Auto-Detection

🧠 Smart Dev Server & Terminal Timeout System

• Background Dev Server Auto-Detection: When running continuous development servers (such as npm run dev, vite, npm start, or local web servers), Yengi no longer gets stuck or hangs waiting for process exit.
• Smart 15s Timeout: Terminal commands now use an intelligent 15-second startup detection window. If a background server is detected (e.g. http://localhost, listening, compiled), Yengi marks it as [BACKGROUND DEV SERVER ACTIVE] and lets the AI continue its workflow seamlessly.
• Improved Terminal Stability: Short-lived commands exit normally while long-running background servers are handled without blocking the agent.

---

Built with .NET 10 (LTS) | WPF | 201/201 Passing Unit Tests

v1.12 — Interactive Tool Permissions & Self-Correction

🛡️ Interactive Tool Permissions & Self-Correction (v1.12)

1. User Permission Gate for Unselected Tool Requests

When the AI requests a tool not pre-selected by the Router, Yengi now prompts for explicit user permission:

• [ ✅ Allow & Add Tool ] — Dynamically adds the tool to the catalog and lets AI continue safely.
• [ ❌ Deny Permission ] — Politely notifies the AI to proceed without using that tool.

2. Self-Correction Retry Loop for Missing Tool Arguments

If a tool invocation fails due to missing JSON arguments, Yengi automatically sends a self-correction hint back to the model:

• The AI auto-corrects its parameters on the next turn without stopping the workflow or confusing the user.

3. Loop Continuation Fix

Resolved an issue where approving "Continue 10 More Steps" did not re-enter the loop body.

---

Built with .NET 10 (LTS) | WPF | Full TR/EN/ZH Localization

v1.11 — AI Continuation Loop Fix, Informative Feature Popups & Smart Persistence

🐛 Critical Fixes

• AI Continuation Loop Repair: Fixed an issue where clicking "Continue 10 More Steps" did not re-enter the tool execution loop. The AI now seamlessly continues analyzing files beyond step 10.

💡 Feature Popups & Smart Memory

• Research Mode & Image Studio Popups: Added informative modal guides with "Do Not Show Again" persistent settings for capability toggles.
• Smart Workspace Memory: Automatically restores your last opened project folder on launch.

---

Built with .NET 10 (LTS) | WPF

v1.10 — Informative Feature Popups, Smart Persistence & UI Refinements

💡 Informative Capability Popups

Added elegant, dark-themed informative modal dialogs for main AI feature toggles:

• 🔎 Research & Chat Mode Info: Explains that coding agents, terminal commands, and build verification steps are bypassed in this mode to focus purely on web research, brainstorming, Q&A, and .md/.txt note generation.
• 🎨 AI Image Generation Info: Explains how autonomous image generation works and reminds users that keeping it disabled when not needed saves API quota and speeds up response times.
• "Do Not Show Again" Option: Both popups feature a persistent [x] Do not show again checkbox backed by application settings.

🔎 Research & Chat Mode

• Live web search powered by Tavily API with free DuckDuckGo search fallback
• Autonomous Markdown (.md) research reports with embedded AI-generated illustrations
• Full localization across English, Turkish, and Chinese (TR/EN/ZH)

💾 Workspace Persistence & UI Polish

• Automatically restores your last opened project folder on every launch
• Added a vertical separator line in the chat toolbar to cleanly separate panel views from capability toggles
• Resolved editor scrollbar overlay issue for seamless drag-and-drop scrolling

---

201 unit tests passing ✅ | Built with .NET 10 (LTS) | WPF

v1.09 — Research & Chat Mode, AI-Illustrated Reports & Smart Project Memory

🔎 Research & Chat Mode

A brand-new AI mode designed for learning, brainstorming, and research.
Toggle the 🔎 button in the chat toolbar to activate it.

• Live web search powered by Tavily API (with free DuckDuckGo fallback — no API key required)
• Automatically saves research results as clean, structured Markdown (.md) reports to your project folder
• Supports freeform chat & brainstorming — ask questions, develop ideas, take notes

🎨 AI-Illustrated Research Reports

Activate both 🔎 Research and 🎨 Image Generation toggles simultaneously to get:

• Live web research + automatic AI-generated illustrations embedded directly into the .md report
• One prompt → a complete, illustrated research document ready to share or print