AI-powered academic abstract generator for medical imaging and scientific research conferences.

English | 中文 | Deutsch

Sci-Necromancer streamlines the process of preparing conference abstracts by leveraging large language models to analyze research content, generate impact statements, suggest appropriate submission categories, and produce polished abstracts that conform to conference guidelines.

Migration Status: This project has been migrated from React 19 to Vue 3 + Composition API with a Nest.js serverless backend for file processing.

• ISMRM - International Society for Magnetic Resonance in Medicine
• RSNA - Radiological Society of North America
• ESC - European Society of Cardiology
• ECR - European Congress of Radiology

• Standard Analysis Mode: Upload PDF/DOCX or paste text, then follow a guided workflow:
  • Content analysis with keyword extraction
  • Impact statement and synopsis generation
  • Abstract type suggestion based on content
  • Final abstract generation conforming to conference guidelines
• Creative Expansion Mode: Provide a core research idea and generate a complete abstract directly

• Standard Edit: Upload images and apply AI-powered editing based on specifications
• Creative Generation: Generate figures from abstract context (impact + synopsis)

• Save and organize generated abstracts locally
• Import/export abstracts as JSON for backup and sharing
• Full CRUD operations on saved abstracts

• Markdown (.md)
• PDF document
• JSON data

• English and Chinese language support
• Automatic browser language detection
• Easy language switching via UI

• Node.js 18 or higher
• API key for Google AI (Gemini) or an OpenAI-compatible provider

# Clone the repository
git clone https://github.com/yourusername/sci-necromancer.git
cd sci-necromancer

# Install dependencies
npm install

# Start development server
npm run dev

The application will be available at http://localhost:3000.

npm run build
npm run preview

1. Launch the application and click the Models button (gear icon) in the header
2. Select your AI provider:
   • Google AI: Enter your Gemini API key from Google AI Studio
   • OpenAI-compatible: Enter base URL and API key (supports OpenAI, SiliconFlow, and other compatible APIs)
3. Configure models (optional):
   • Text model (e.g., gemini-2.5-pro, gpt-4o)
   • Vision model (for image analysis)
   • Image model (for figure generation)
4. Save settings

Settings are stored in browser localStorage and persist across sessions.

1. Select Conference: Choose the target conference tab (ISMRM, RSNA, ESC, ECR)
2. Input Content: Upload a PDF/DOCX file or paste your research text
3. Analyze: Click "Analyze" to extract categories and keywords
4. Generate Impact & Synopsis: Create impact statement and synopsis
5. Select Type: Review AI-suggested abstract types or choose manually
6. Generate Abstract: Produce the final abstract conforming to conference guidelines
7. Export: Download as Markdown, PDF, or JSON

1. Select a conference and switch to "Creative Expansion" mode
2. Enter your core research idea or hypothesis
3. Generate a complete abstract directly from the concept

1. Navigate to the "Figure Generation" tab
2. Choose mode:
   • Standard Edit: Upload an image and specify editing instructions
   • Creative Generation: Generate figures based on your abstract context
3. Download the generated figure

The ECR module integrates EQUATOR Network reporting guidelines to ensure your abstract follows best practices for your study type:

For more information, visit the EQUATOR Network.

• Research Presentation (RP) - 5-minute oral presentation
• Clinical Trials in Radiology (CTiR) - 8-minute presentation for multicentre/randomised studies
• EPOS Scientific Poster - Electronic poster with scientific content
• EPOS Educational Poster - Teaching-focused electronic poster
• Student Presentation - For university projects and first research papers

ECR Abstract Submission: www.myESR.org/abstractsubmission

sci-necromancer/
├── src/
│   ├── main.ts                 # Application entry point
│   ├── App.vue                 # Root component with error boundary
│   ├── components/
│   │   ├── panels/             # Conference-specific panels (ISMRM, RSNA, ESC, ECR)
│   │   │   ├── ConferencePanel.vue
│   │   │   ├── ISMRMPanel.vue
│   │   │   ├── RSNAPanel.vue
│   │   │   ├── ESCPanel.vue
│   │   │   ├── ERPanel.vue
│   │   │   └── ImageGenerationPanel/
│   │   ├── managers/           # Abstract and Model configuration
│   │   ├── ui/                 # Reusable UI components
│   │   └── export/             # Export functionality
│   ├── composables/            # Vue composables (useLLM, useSettings, etc.)
│   ├── plugins/                # Vue plugins (i18n, errorHandler)
│   └── services/               # Frontend services
├── lib/
│   ├── llm/                    # LLM provider integrations
│   │   ├── index.ts            # Unified API facade
│   │   ├── gemini.ts           # Google AI integration
│   │   ├── openai.ts           # OpenAI-compatible integration
│   │   └── prompts/            # Conference-specific prompts
│   ├── conference/             # Conference module system
│   │   ├── modules/            # Per-conference implementations
│   │   ├── BaseConferenceModule.ts
│   │   ├── ConferenceRegistry.ts
│   │   └── ConferenceRouter.ts
│   ├── file/                   # File processing utilities
│   ├── hooks/                  # Shared hooks (theme, keyboard nav)
│   ├── i18n/                   # Internationalization config
│   └── utils/                  # Shared utilities
├── server/                     # Nest.js serverless backend
│   └── src/
│       ├── main.ts             # Serverless entry point
│       ├── app.module.ts       # Root module
│       └── file/               # File processing module
│           ├── file.controller.ts
│           ├── file.service.ts
│           └── dto/
├── services/                   # Legacy services (being migrated)
├── public/
│   ├── locales/                # i18n translation files
│   └── *.md                    # Conference guidelines
├── types.ts                    # TypeScript type definitions
├── vite.config.ts              # Vite configuration
└── tsconfig.json               # TypeScript configuration

The Nest.js backend is located in the server/ directory:

cd server
npm install
npm run start:dev

The @ alias maps to the project root:

import { useSettings } from '@/src/composables/useSettings';
import { analyzeContent } from '@/lib/llm';

1. Create a new module in lib/conference/modules/ extending BaseConferenceModule
2. Define conference-specific guidelines, types, and prompts
3. Register the module in ConferenceRegistry
4. Create a corresponding panel component in src/components/panels/
5. Add translations in public/locales/

• State Management: Uses Pinia stores with pinia-plugin-persistedstate for localStorage persistence
• Error Handling: Global error boundary via onErrorCaptured hook in App.vue
• LLM Integration: Unified facade in lib/llm/index.ts handles provider selection
• File Processing: Heavy parsing (PDF/DOCX) delegated to Nest.js backend, avoiding browser-side overhead

1. Push the repository to GitHub
2. Import the project in Vercel
3. Configure:
   • Framework preset: Vite
   • Build command: npm run build
   • Output directory: dist
4. For serverless backend, configure the server/ directory as a separate Vercel Function
5. Deploy

No environment variables are required; API keys are entered via UI and stored locally.

The Nest.js backend in server/ supports deployment to:

• Vercel Serverless Functions
• AWS Lambda (via @nestjs/platform-aws-lambda)
• Other serverless platforms with appropriate adapters

The frontend build output is a static site in dist/. Deploy to any static hosting service (Netlify, GitHub Pages, Cloudflare Pages, etc.).

• API keys are stored in browser localStorage only
• No server-side storage or transmission of credentials
• File processing delegated to serverless backend (sandboxed execution)
• Avoid uploading sensitive or confidential research data

Sci-Necromancer supports optional MCP (Model Context Protocol) integrations:

• Supabase MCP: Cloud persistence for abstracts and settings (configure via Model Manager)
• Image Generation MCP: Alternative image generation via MCP-enabled tools
• Chrome DevTools MCP: UI comparison during development

See MCP_TOOLS_GUIDE.md for detailed configuration.

1. Fork the repository
2. Create a feature branch (git checkout -b feature/your-feature)
3. Commit changes (git commit -m 'Add your feature')
4. Push to branch (git push origin feature/your-feature)
5. Open a Pull Request

Please run npm run lint before submitting PRs to ensure type safety.

• README.md - This file (quick start and overview)
• QUICK_REFERENCE.md - Common user tasks and shortcuts
• WORKFLOW.md - Overall workflow documentation
• MODEL_CONFIGURATION_GUIDE.md - Provider setup details
• MCP_TOOLS_GUIDE.md - MCP integration guide
• IMAGE_GENERATION_ARCHITECTURE.md - Figure generation technical details
• lib/llm/WRITING_STYLE_GUIDE.md - Writing style enforcement guidance

• ISMRM, RSNA, ESC, ESR for public abstract guidelines
• Google AI (Gemini) for language model capabilities
• OpenAI for API compatibility standards
• SiliconFlow for image generation APIs
• EQUATOR Network for research reporting guidelines
• Vue.js and Pinia for the frontend framework
• Nest.js for the serverless backend framework

MIT

English | 中文 | Deutsch