1.2.1 - Many bug fixes and enhancements

🌟 What's New

🎨 Visual & Design

• Complete Light Mode Redesign - Warm cream backgrounds with purple/pink accents
• Enhanced Dark Mode - Improved crystalline effects and better contrast
• Holographic Shard Design - Sharp angular borders and neon glow effects
• Modern Icons - Replaced brain emoji with AI-appropriate Sparkles icon

🔧 Technical Improvements

• 9 New Reusable UI Components - IconButton, GlassPanel, LoadingCard, and more
• Modular Architecture - Reorganized search suggestions into self-contained module
• Enhanced Error Handling - Clean, contextual error messages without UI clutter
• Performance Optimized - Maintained 71% build time improvement (2.0s builds)

🌐 Browser & Platform Support

• Universal Compatibility - Consistent experience across Firefox, Chrome, Safari
• IPv6 & Docker Support - Full IPv6 networking for modern deployments
• Cross-Browser Fixes - Resolved theme switching and rendering issues

🚀 User Experience

• Improved AI Responses - Better markdown formatting and text stability
• Enhanced Search Interface - Restored feature panels with crystalline styling
• Better Loading States - Improved skeleton components and progress indicators
• PDF Generation - Optimized layout with focus on AI content

---

🐛 Key Fixes

• Hydration Mismatch Errors - Resolved React SSR inconsistencies
• Chrome Dark Mode Issues - Fixed card backgrounds and text visibility
• Text Readjustment - Eliminated visual jumps in AI response cards
• CSS Conflicts - Resolved Tailwind utility and custom style conflicts

---

📦 Deployment & Configuration

• Docker IPv6 Support - Enhanced container networking
• Environment Setup - Improved API endpoint configuration
• Build Optimization - Streamlined development and production builds
• Zero Breaking Changes - Full backward compatibility with v1.2.0

---

📊 Performance Metrics

• ✅ 2.0s Build Time (71% improvement maintained)
• ✅ 100% Cross-Browser Compatibility
• ✅ 13 Files Cleaned Up (50KB+ space saved)
• ✅ 115+ Lines of Duplicate Code Removed

---

🔄 Migration

No action required - All improvements are automatically applied. For best experience:

1. Clear browser cache (Ctrl+Shift+R)
2. Update Docker IPv6 settings if using containers
3. Verify environment variable configuration

---

📚 Resources

• Full Release Notes: RELEASE_NOTES_v1.2.1.md
• Documentation: README.md
• Issues: GitHub Issues

---

Happy Searching! 🔍✨

1.2.0 - Major overhaul

Cognito AI Search v1.2.0 · May 30, 2025

🚀 Massive Redesign & UX Overhaul

• New Holographic Shard Design System
• Sleek, crystalline UI with dark/light mode and mobile support
• Gradient typography, neon-glow elements, and smooth transitions

⚡ Performance Gains

• Build time: 5s → 2s (60% faster)
• Removed 30K+ lines of unused UI code and 28 dependencies
• Smaller bundle, faster load, leaner deploys

🧠 Enhanced Features

• 200+ smart AI suggestions across 16 categories
• Pro-level PDF export (Markdown, LaTeX, syntax highlighting)
• Intelligent caching, deduplication, and expiration

🔧 Code Quality

• Modular structure with reusable components
• Full TypeScript coverage
• API separation (Ollama & SearXNG layers)

🐛 Fixes

• Light mode styling bugs fixed
• Firefox rendering issues resolved
• Ollama 0.9.0+ compatibility ensured

📄 See full notes in docs/RELEASE_NOTES_v1.2.0.md

1.1.0 - Major improvements

🚀 Cognito AI Search v1.1.0

A major release introducing LaTeX math rendering, sophisticated animations, and comprehensive codebase optimizations that transform Cognito AI Search into a polished, production-ready application.

---

✨ What's New

🧮 LaTeX Math Rendering Support

Mathematical expressions in AI responses now render beautifully! Support for both inline ($E = mc^2$) and block math ($$\int_{-\infty}^{\infty} e^{-x^2} dx = \sqrt{\pi}$$) using KaTeX.

🎨 Enhanced UI/UX

• Modern AI Icon: Replaced brain emoji with sleek Sparkles icon
• Smooth Animations: 60fps transitions with coordinated timing
• Professional Polish: Eliminated text readjustment and jarring state changes

⚡ Performance Optimizations

• 49% CSS Reduction: Streamlined from 481 to 244 lines
• 19 Files Removed: Eliminated unused components and duplicates
• 2.0s Build Time: Maintained 71% faster builds with zero breaking changes

🌐 Network & Infrastructure

• IPv6 Support: Future-proof network connectivity
• Enhanced Docker: Updated to Node.js 24-alpine
• Improved Configuration: Better environment variable handling

---

📊 Key Metrics

Metric	Improvement
Build Performance	71% faster (2.0s)
CSS Size	49% reduction
Files Cleaned	19 unused files removed
Breaking Changes	Zero
Animation Performance	60fps smooth transitions

---

🔧 Technical Highlights

• Math Rendering: remark-math + rehype-katex + KaTeX
• Modern Stack: Full Tailwind CSS v4 adoption
• Clean Architecture: Eliminated duplicate and legacy code
• Robust Error Handling: Better user feedback and troubleshooting

---

📖 Full Release Notes

For comprehensive details, technical specifications, migration notes, and complete changelog, see:
📋 Complete Release Notes

---

🎯 What's Next

This release establishes Cognito AI Search as a professional, production-ready AI search platform. The foundation is now set for advanced features and integrations in future releases.

Try the new math rendering by asking mathematical questions! 🧮✨

---

Questions? Check our documentation or open an issue.

1.0.6 - Docker finally working

I had to go a few rounds to figure out the best way to get everything working both on bare metal and in Docker and it has now been fixed.

What's Changed

• chore(deps)(deps): bump node from 20-alpine to 24-alpine by @dependabot in #7
• Redesign AI response loading component

New Contributors

• @dependabot made their first contribution in #7

Full Changelog: 1.0.5...1.0.6

1.0.5 - Urgent Hotfix

This is an urgent hotfix that fixes the issue with the main page not displaying correctly. This was due to overlapping config options in the backend which hindered the frontend from getting information successfully from Ollama and SearXNG.

Full Changelog: 1.0.4...1.0.5

1.0.4 - Fix Docker build issue

pnpm could, sometimes, crash and consume all available memory. This bug-fix update fixes this by limiting the available memory to 4GB.

Full Changelog: 1.0.3...1.0.4

1.0.3 - Bugfix release

This is a tiny bug-fix release where the max token was set to 100 instead of 1200, cutting the response short.
The next bug was that the fallback model was used instead of the one defined in the environment file.

1.0.2 - Docker support

Cognito AI Search v1.0.2 Release Notes

We're excited to announce the release of Cognito AI Search version 1.0.2!

This release introduces comprehensive Docker support, making it easier than ever to deploy and manage Cognito AI Search in containerized environments. We've also included important updates to ensure consistency between local development and production standalone builds.

---

✨ New Features & Enhancements

• Full Docker Support (Commit fd61147):

  • Dockerfile: A multi-stage Dockerfile has been added, utilizing pnpm and node:20-alpine for optimized, lightweight, and secure production images.
  • .dockerignore: Included to ensure a lean build context, reducing image size and build times.
  • Standalone Output: next.config.mjs now uses output: 'standalone' for optimized Next.js builds tailored for containerization.
  • Updated Documentation: The HOWTO.md file has been significantly updated with a new "Docker Deployment" section. This includes:
    • Instructions for building the Docker image (including passing build arguments).
    • Guidance on running the Docker container with necessary environment variables.
    • An example docker-compose.yml for easy integration with SearXNG and optional Ollama services.
  • A minor linting fix was applied to the Dockerfile for comment formatting.

• Improved Standalone Start Script (Commit e4cded2):

  • The start script in package.json has been updated to HOSTNAME=0.0.0.0 node .next/standalone/server.js.
  • This change ensures that when running the application locally after a build (pnpm build && pnpm start), it correctly uses the standalone server script and listens on all network interfaces (0.0.0.0). This provides consistent behavior with Docker deployments and makes the application reliably accessible via localhost and other network IPs.

---

⬆️ Version Update

• The application version in package.json has been bumped to 1.0.2.

---

🚀 Getting Started

To take advantage of these new features, pull the latest changes from the main branch. If you're planning to deploy with Docker, please refer to the updated "Docker Deployment" section in HOWTO.md.

We believe these enhancements will significantly improve the deployment flexibility and development experience for Cognito AI Search. As always, we welcome your feedback and contributions!

1.0.1 - Minor bug fixes

Fixed a few minor bugs in the code and updated the documentation.

• Fixed issue where the optimized search string was sent to Ollama instead of the original
• Updated How it works with correct information
• Cleaned up and updated all the documentation

1.0.0 - Cognito AI Search

I'm pleased to announce the release of version 1.0.0 of Cognito AI Search.

I'm very pleased with the results so far and I really hope that you will see something here that helps you. It's been a few very hectic and long days vibe coding it in Windsurf. These last steps was done with the help of Claude Sonnet 4 and that means I've spent quite a bit of money perfecting it, or at least made it as good as possible for version 1.0.0.

Enjoy and if you have an issue or a suggestion, either create a PR or a new Issue.