- P4.0.4 Asset Linking Engine — citation→Figure/Table/Supplementary/Dataset links
- P4.1 Figure Intelligence — Figure Cards with 14 evidence types + quality scoring
- P4.2 Table Intelligence — xlsx/csv/tsv parsing, 17 table types, statistical field detection
- P4.3 Supplementary Intelligence — PDF/DOCX/TXT parsing, section detection, evidence extraction
- P5.1 Unified Evidence Graph — 14 node types + 22 edge types, GraphML/Cytoscape.js export
- P5.2 Cross-Asset Query Engine — unified query across all asset types
- P5.3 Dataset Intelligence — entity extraction, numeric profiling, cross-paper comparison
- P5.4 Graph-Augmented Research Agent — 8-tool agent with evidence chains + research plans
- P6.0 End-to-End Validation — system audit, paper status matrix
- P6.1 Quality Dashboard — pipeline health, paper quality table, P0-P3 recommendations
- P6.2 Demo Project — synthetic demo dataset
- P6.3 Documentation & Workflow Polish — CLI/API/troubleshooting docs
- P6.4 CI/CD & Release Hardening — gitignore, secret scan, GitHub Actions
- LLM Gateway — unified DeepSeek/OpenAI/Anthropic/Local interface
- Summary V2 + Evidence Enrichment
- Gap Extraction + Hypothesis Generation
- Cross-Paper Gap/Hypothesis Fusion
- Research Opportunity Ranking
- Hybrid PDF Parser