graphrag-expert

What problem does it solve?

GraphRAG teams need a reliable way to retrieve, ground, and cite knowledge-graph evidence when answering biomedical questions for drug repurposing and mechanism-of-action discovery.

Core Features & Use Cases

• Hybrid retrieval design: combines Cypher-based structured querying with vector similarity search to identify relevant entities and paths in the OSPF Ayurveda knowledge graph.
• Evidence-preserving context assembly: turns 1–3 hop subgraph traversal results into grounded LLM context while preserving provenance fields like pmid, gdaScore, and source.
• Prompt engineering and Cypher generation guardrails: crafts prompts that include the full Neo4j schema and enforce non-hallucination and citation requirements.

Quick Start

Use the graphrag-expert skill to draft an intent-classifier plus hybrid retrieval plan (vector entry discovery, Cypher generation, 1–3 hop traversal, and citations) for a specific Oral Mucositis question.