Graph RAG pipeline (three tiers, RRF fusion)

Generated from mcp/patterns.registry.json — do not edit by hand; change the registry and run bun run patterns:build.

Index card — the docs below are the canonical explanation; deep-tier pattern pages additionally carry invariants and emulation notes.

Category: AI · Tier: light · Risk: medium — external LLM providers with quota limits

Vector-similarity chunk retrieval is combined with Neo4j knowledge-graph traversal across three parallel tiers, fused by reciprocal rank fusion, for more explainable answers.

When to use: Use when vector search alone misses multi-hop relationship queries that a knowledge graph can answer.

Docs

• docs/blueprint/ai/graph-rag.md (GitHub · GitLab)

Code

• src/lib/server/rawrag/tiers/ (GitHub · GitLab)
• src/lib/server/graph/rag/ (GitHub · GitLab)

Proof

• /showcases/ai/retrieval/rag-chat

Machine-readable record: ai-graph-rag in mcp/patterns.registry.json.