Modules Overview

Omelette is organized into modular components that form a complete research literature pipeline.

Core Pipeline

#	Module	Description
1	Keywords	Three-level hierarchy, LLM expansion, search formula
2	Search	Federated search across multiple academic sources
3	Deduplication	DOI hard match → title similarity → LLM verify
4	Subscription	RSS and API-based scheduled updates
5	Crawler	PDF download via Unpaywall, arXiv, direct URL
6	OCR	MinerU + pdfplumber + PaddleOCR
7	RAG	LlamaIndex + ChromaDB, hybrid retrieval
8	Writing	Summarization, citations, gap analysis

Pipeline Flow

Keywords → Search → Dedup → Subscription → Crawler → OCR → RAG → Writing

Each module can be used independently or as part of the full pipeline.

Extended Modules

Beyond the core pipeline, Omelette includes additional modules for research management:

• Analysis: Trend analysis, author networks, gap analysis, impact scoring
• Collections: Custom paper groups with AI-suggested tags
• Concepts: LLM-powered concept extraction and knowledge graph
• Chat: Streaming conversational interface with tool modes
• Export: BibTeX, RIS, EndNote, Zotero export
• Library: Metadata health check and repair
• Notifications: In-app alerts for subscription matches
• Team Members: Project collaboration with RBAC
• Pipelines: LangGraph orchestration with HITL interrupt/resume
• Audio Overviews: LLM-generated paper discussion audio

See the API Reference for complete endpoint documentation.