Do I need a separate graph database like Neo4j?

No, this skill enables graph-like capabilities within PostgreSQL by utilizing tables for entities and edges, combined with vector and full-text search extensions.

What search methodologies does it support?

It generates SQL for Local Search (n-hop entity expansion), Global Search (community-level summaries), and Hybrid Search (combining vector and keyword results).

How does it handle data extraction?

The skill generates custom LLM prompts (entity_extraction.md and edge_extraction.md) tailored to your specific schema to help automate the population of your graph.

Does it support languages other than English?

Yes, it includes specific indexing strategies for Japanese and English using PGroonga's TokenMecab and TokenBigram tokenizers.

What is the primary purpose of the GraphRAG PostgreSQL skill?

It allows developers to build relationship-aware Retrieval-Augmented Generation (RAG) systems directly within PostgreSQL by automating schema and query generation.

GraphRAG for PostgreSQL

Name: GraphRAG for PostgreSQL
Author: titabash

bytitabash

Database Management

Automates the design and implementation of GraphRAG systems using PostgreSQL, pgvector, and PGroonga.

About

This skill transforms natural language project descriptions into production-ready GraphRAG architectures within a PostgreSQL environment. It leverages pgvector for semantic search and PGroonga for advanced full-text indexing, automatically generating entity/edge schemas, specialized query patterns for local and global search, and LLM-driven data extraction prompts. It is ideal for developers who want to build sophisticated, relationship-aware RAG systems without leaving the PostgreSQL ecosystem.

Key Features

0 GitHub stars
Support for advanced features like time-series relationship tracking and community detection
Domain-specific LLM prompts for automated entity and relationship extraction
Automated entity and relationship (edge) schema design based on natural language requirements
Generation of specialized SQL patterns for Local, Global, and Hybrid search queries
Integration of pgvector for semantic vector search and PGroonga for hybrid search capabilities

Use Cases

Building a relational knowledge base for complex technical documentation and API references
Managing evolving character relationships and world-building elements for creative storytelling
Implementing advanced RAG for troubleshooting logs to perform root cause and impact analysis

About

Key Features

0 GitHub stars
Support for advanced features like time-series relationship tracking and community detection
Domain-specific LLM prompts for automated entity and relationship extraction
Automated entity and relationship (edge) schema design based on natural language requirements
Generation of specialized SQL patterns for Local, Global, and Hybrid search queries
Integration of pgvector for semantic vector search and PGroonga for hybrid search capabilities

Use Cases

Building a relational knowledge base for complex technical documentation and API references
Managing evolving character relationships and world-building elements for creative storytelling
Implementing advanced RAG for troubleshooting logs to perform root cause and impact analysis