todo.md

Helix Creative Studio - Project TODO

Phase 1: Core Features

• Database schema for stories, UCF states, and agent logs
• Story generation API with Z-88 ritual engine integration
• Real-time UCF visualization dashboard
• Agent status monitoring interface
• Story archive browser with search/filter
• Story detail view with metadata display

Phase 2: UI/UX

• Landing page with cyberpunk aesthetic
• Story generation form with prompt input
• Real-time generation progress indicator
• UCF metrics visualization (charts/graphs)
• Agent activity timeline
• Story card grid layout
• Responsive design for mobile/tablet

Phase 3: Advanced Features

• User authentication and story ownership
• Story editing and regeneration
• Export stories (Markdown, PDF)
• Share stories with public links
• Story collections/anthologies
• Advanced filtering (by quality, date, agents used)

Phase 6: Quality of Life Improvements

• Database: Add collections/folders table
• Database: Add tags support
• Database: Add favorites/bookmarks
• Database: Add soft delete (deletedAt field)
• Feature: Delete single story
• Feature: Bulk delete stories
• Feature: Create and manage collections
• Feature: Organize stories into collections
• Feature: Search stories by title/content
• Feature: Filter by collection, tags, favorites
• Feature: Sort by date, quality, word count
• Feature: Add/remove tags
• Feature: Favorite/unfavorite stories
• Feature: Bulk export stories as ZIP
• UI: Delete confirmation dialog
• UI: Collection management interface
• UI: Search bar in archive
• UI: Filter dropdowns
• UI: Sort controls
• UI: Tag editor
• UI: Bulk selection checkboxes

Phase 5: Multi-LLM Enhancement

• Backend: LLM router supporting OpenAI, Anthropic, xAI, Google, Perplexity
• Backend: Agent→LLM mapping system with defaults
• Backend: 7 agents configured (Oracle, Lumina, Gemini, Agni, Claude, Kavach, Researcher)
• Backend: 5 preset modes (Balanced, Creative, Structured, Experimental, Research-Grounded)
• Backend: Multi-LLM Z-88 engine (z88EngineMulti.ts)
• Backend: API endpoints for agent/preset configuration
• Backend: Streaming support for real-time progress
• UI: Simple preset modes (Balanced, Creative, Structured, Experimental)
• UI: Advanced Options panel (collapsible)
• UI: Per-agent LLM selection dropdowns
• UI: Agent multiplicity controls (1-4x per agent)
• UI: Temperature/creativity sliders per agent
• UI: AgentConfigurator component with full configuration
• UI: Real-time configuration display
• Integration: AgentConfigurator integrated into Generate page
• Feature: Ensemble Mode (parallel generation + synthesis)
• Feature: Save/load custom agent configurations
• Integration: Connect Generate mutation to multi-LLM engine
• Feature: Auto-condense/enhance prompts for better story generation
• Feature: Continuous story chapters (reference archived stories)
• Feature: Story series/collections linking
• Feature: "Continue Story" button on story detail pages
• Database: Series support (seriesId, chapterNumber, previousChapterId)
• Backend: Story continuation prompt generation
• Bug: Loading bar not showing during generation (fixed with multi-LLM integration)

Bug Fixes

• Fix "Story Not Found" routing issue (use ritualId instead of numeric id)
• Increase prompt character limit from 500 to 1000
• Add getByRitualId API endpoint for proper story routing
• Improve loading bar accuracy (better progress distribution, goes to 98% before completion)

Phase 4: Deployment

• Production build optimization
• Database migrations
• Environment variable documentation
• GitHub repository created and pushed
• Comprehensive README with setup instructions
• Deployment documentation (Manus, Railway, Docker)
• Contributing guidelines for collaboration
• Railway deployment (optional)
• Docker containerization (optional)
• Post-deployment verification