GraphRAG Toolkit
AWS GraphRAG Toolkit is an open-source framework for building knowledge graph applications with Large Language Models (LLMs). FalkorDB is supported as a graph store backend, enabling you to leverage FalkorDB’s high-performance graph database capabilities in your GraphRAG applications.
Overview
The GraphRAG Toolkit provides tools and patterns for building retrieval-augmented generation (RAG) applications that use knowledge graphs. With FalkorDB as the graph store, you can:
- Build and query knowledge graphs efficiently
- Use semantic-guided search for intelligent retrieval
- Connect to FalkorDB Cloud or local instances
- Integrate with LLM-powered applications
Installation
The FalkorDB graph store is contained in a separate contributor package. Install it using:
pip install https://github.com/awslabs/graphrag-toolkit/archive/refs/tags/v3.13.3.zip#subdirectory=lexical-graph-contrib/falkordb
Quick Start
1. Register FalkorDB as a Graph Store
Before creating a FalkorDB graph store, register the FalkorDBGraphStoreFactory with the GraphStoreFactory:
from graphrag_toolkit.lexical_graph.storage import GraphStoreFactory
from graphrag_toolkit_contrib.lexical_graph.storage.graph.falkordb import FalkorDBGraphStoreFactory
GraphStoreFactory.register(FalkorDBGraphStoreFactory)
2. Create a FalkorDB Graph Store
You can use the GraphStoreFactory.for_graph_store() static factory method to create an instance of a FalkorDB graph store.
Using FalkorDB Cloud
To create a FalkorDB Cloud graph store, supply a connection string that begins with falkordb://, followed by your FalkorDB endpoint:
from graphrag_toolkit.lexical_graph.storage import GraphStoreFactory
from graphrag_toolkit_contrib.lexical_graph.storage.graph.falkordb import FalkorDBGraphStoreFactory
falkordb_connection_info = 'falkordb://your-falkordb-endpoint'
GraphStoreFactory.register(FalkorDBGraphStoreFactory)
with GraphStoreFactory.for_graph_store(falkordb_connection_info) as graph_store:
# Your code here
pass
You may need to pass credentials and SSL configuration:
from graphrag_toolkit.lexical_graph.storage import GraphStoreFactory
from graphrag_toolkit_contrib.lexical_graph.storage.graph.falkordb import FalkorDBGraphStoreFactory
falkordb_connection_info = 'falkordb://<your-falkordb-endpoint>'
GraphStoreFactory.register(FalkorDBGraphStoreFactory)
with GraphStoreFactory.for_graph_store(
falkordb_connection_info,
username='<username>',
password='<password>',
ssl=True
) as graph_store:
# Your code here
pass
Using Local FalkorDB
To create a local FalkorDB graph store, supply a connection string with only falkordb://:
from graphrag_toolkit.lexical_graph.storage import GraphStoreFactory
from graphrag_toolkit_contrib.lexical_graph.storage.graph.falkordb import FalkorDBGraphStoreFactory
falkordb_connection_info = 'falkordb://'
GraphStoreFactory.register(FalkorDBGraphStoreFactory)
with GraphStoreFactory.for_graph_store(falkordb_connection_info) as graph_store:
# Your code here
pass
3. Start FalkorDB (Local Setup)
If you’re using a local instance, start FalkorDB with Docker:
docker run -p 6379:6379 -p 3000:3000 -it --rm falkordb/falkordb:edge
Or sign up for FalkorDB Cloud for a managed solution.
Features
Semantic-Guided Search
The FalkorDB graph store supports semantic-guided search, enabling intelligent retrieval based on meaning and context rather than just keyword matching.
Note: The FalkorDB graph store currently does not support traversal-based search.
Resources
- 🔗 AWS GraphRAG Toolkit GitHub Repository
- 📖 GraphRAG Toolkit Documentation
- 📓 FalkorDB Graph Store Documentation
- ☁️ FalkorDB Cloud
Use Cases
- Knowledge Graph Construction: Build structured knowledge graphs from unstructured data
- Semantic Search: Implement context-aware search using graph-based retrieval
- Question Answering: Combine LLMs with graph data for accurate responses
- Document Understanding: Extract and organize information in a knowledge graph
Related Tools
- GraphRAG-SDK: FalkorDB’s native GraphRAG solution
- LangChain: Build AI agents with graph memory
- LlamaIndex: LLM application framework with FalkorDB support