An interactive web explorer that presents interconnected articles with AI-generated content. Navigate through mysterious topics, uncover hidden connections, and explore knowledge through multiple writing perspectives.
- Dynamic Article Creation: Powered by OpenAI-compatible APIs for endless content
- Enhanced 3-Phase Generation: Extract facts → Generate relationships → Synthesize article
- Smart Caching: Articles cached for 7 days to minimize API costs
- Progressive Enhancement: Works without API key in demo mode
Choose from 6 distinct AI personalities:
- Dramatic: Compelling narratives with dramatic flair
- Academic: Scholarly analysis and rigorous exploration
- Mystical: Esoteric wisdom and occult knowledge
- Investigative: Forensic journalism and fact-finding
- Poetic: Lyrical storytelling with metaphorical language
- Scientific: Rigorous precision with data-driven analysis
- Clickable Keywords: All proper names, places, and concepts are explorable
- Contextual Navigation: "Dig deeper", "Trivia", "Mythology" buttons adapt to current content
- Breadcrumb History: Track your exploration journey
- Favorites System: Save interesting articles for later reference
- Modal Search Interface: Dedicated search experience with quick topics
- Popular Topics: One-click access to fascinating subjects
- Keyboard Shortcuts:
Cmd/Ctrl + Kfor instant search
- Developer Mode: Advanced tools and debugging (auto-activates on localhost)
- Keyboard Shortcuts: Full hotkey support for power users
- Real-time Configuration: Hot-reload prompts and settings
- Cache Management: Inspect and clear article cache
- Modern web browser (Chrome, Firefox, Safari, Edge)
- OpenAI API key (optional, for dynamic content generation)
- Clone the repository:
git clone [email protected]:dotcommander/scrolls.git
cd scrolls- Configure API access (optional):
cp public/config.local.js.example public/config.local.js
# Edit config.local.js with your API key- Open in browser:
- Direct file: Open
public/index.htmlin your browser - Local server: Serve the
public/directory with any web server
- Direct file: Open
- Explore Topics: Click any highlighted keyword to dive deeper
- Navigate History: Use breadcrumbs to retrace your journey
- Save Favorites: Click the heart icon on articles you want to keep
- Search: Use the search icon or
Cmd/Ctrl + Kfor quick access
Click the model name in the footer to cycle through available models:
gpt-4.1: Advanced reasoning and creativitygpt-4o: Optimized performance and speedgpt-4o-mini: Cost-effective optiongpt-4.1-mini: Balanced performance
Click the writing style button to choose your preferred AI personality:
- Each style generates completely different perspectives on the same topics
- Styles affect tone, complexity, and approach to the subject matter
- Try different styles on the same topic for varied insights
Activation: Automatically on localhost or add ?dev=true to any URL
Keyboard Shortcuts:
Cmd/Ctrl + .- Toggle developer panelCmd/Ctrl + K- Quick search modalCmd/Ctrl + Shift + C- Clear all caches1-6- Quick style switching (Dramatic, Academic, Mystical, Investigative, Poetic, Scientific)R- Reload prompt configuration
- Click the settings icon (⚙️) in the header
- Enter your OpenAI API key
- Optionally configure a custom endpoint for:
- Local models (Ollama, etc.)
- Corporate proxies
- Alternative AI providers
- Test connection before saving
The application supports extensible prompt configuration:
- Edit prompts configuration:
public/prompts/config.json - Create style files: Add
.txtfiles inpublic/prompts/styles/ - Automatic detection: New styles appear in the UI automatically
Enable 3-phase generation in settings for richer, more connected content:
- Phase 1: Extract key facts and entities
- Phase 2: Generate surprising connections
- Phase 3: Synthesize enhanced article with deeper insights
scrolls/
├── public/
│ ├── index.html # Main application (single-file SPA)
│ ├── config.local.js.example # API configuration template
│ └── prompts/
│ ├── config.json # Prompt system configuration
│ └── styles/ # Writing style system prompts
│ ├── academic-system.txt
│ ├── conspiracy-system.txt
│ └── mystical-system.txt
├── CLAUDE.md # AI assistant instructions
├── LICENSE # MIT License
├── README.md # This documentation
└── screenshot.png # Application preview
- Zero Build Process: Direct browser execution, no compilation required
- Framework: Alpine.js for reactive UI, Tailwind CSS for styling
- Architecture: Single-file SPA with modular prompt system
- Storage: Browser localStorage for settings, cache, and favorites
- Typography: Premium web fonts (Playfair Display, Lora, Source Sans Pro)
- Responsive: Mobile-first design with desktop enhancements
- Local-First: All data stored in your browser
- API Keys: Never transmitted except to your configured endpoint
- Private by Default: No telemetry or tracking
- Configurable Endpoints: Support for local AI models
Contributions are welcome! This project follows conventional commit standards:
- Fork the repository
- Create a feature branch (
git checkout -b feature/amazing-feature) - Commit with conventional commits (
feat:,fix:,docs:, etc.) - Push to your branch (
git push origin feature/amazing-feature) - Open a Pull Request
- Follow existing code style and patterns
- Test changes across different browsers
- Update documentation for new features
- Maintain the single-file architecture philosophy
This project is licensed under the MIT License - see the LICENSE file for details.
- Inspiration: Interconnected wiki systems and knowledge exploration tools
- Design Philosophy: Questioning consensus reality through alternative perspectives
- Technical Foundation: Alpine.js, Tailwind CSS, and modern web standards
- Typography: Google Fonts for beautiful, readable text rendering
- Repository: github.com/dotcommander/scrolls
- Issues: Report bugs or request features
- Discussions: Community discussions
Explore. Question. Discover. The scrolls are endless.