Skip to content
Logo CodeGraphContext / Documentation
CodeGraphContext/CodeGraphContext
v0.4.7 3.2k 564
Home › Core Concepts › Database Backends

Database Backends

CodeGraphContext is backend-agnostic. It supports multiple graph database engines to balance performance, ease of use, and scalability.

Comparison Overview

Backend Mode Best For Complexity
KùzuDB Embedded Local development, quick starts Zero Config
FalkorDB Embedded/Remote High-performance semantic queries Low to Medium
Neo4j Server Enterprise-scale, visualization High

1. KùzuDB (Embedded)

KùzuDB is the default recommendation for most users. It is an extremely lightweight, in-process graph database.

  • Zero Infrastructure: No Docker or separate services required.
  • Performance: Optimized for analytical queries (OLAP) on graph data.
  • Storage: Saves the graph as a directory on your local machine (usually in .codegraphcontext/).

Installation: 

pip install real_ladybug

2. FalkorDB (High Performance)

FalkorDB is a low-latency graph database designed for massive throughput.

  • Speed: One of the fastest graph engines available for relationship traversal.
  • Modes:
    • FalkorDB Lite: Runs as an embedded engine (Linux/macOS only).
    • FalkorDB Remote: Connects to a standard Redis-based FalkorDB instance.

Installation (Lite): 

pip install falkordblite

3. Neo4j (Enterprise)

Neo4j is the industry standard for graph databases. Use Neo4j when you need advanced visualization or are working with exceptionally large datasets.

  • Visualization: Use the Neo4j Browser (localhost:7474) to see your code graph in a rich, interactive UI.
  • Scalability: Handles billions of nodes and relationships across distributed environments.

Setup: 1. Start Neo4j via Docker or use AuraDB. 2. Configure CGC to use Neo4j: 

cgc config set-db neo4j
cgc config set-neo4j-uri bolt://localhost:7687

Backend Selection Logic

CGC follows a specific priority when deciding which backend to use:

  1. Explicit: Any database specified via the --database flag or CGC_RUNTIME_DB_TYPE env var.
  2. Configured: The default set via cgc config db.
  3. Automatic:
    • Remote FalkorDB: If FALKORDB_HOST is detected.
    • Unix: Tries FalkorDB Lite → KùzuDB → Neo4j.
    • Windows: Tries KùzuDB → Neo4j.
CodeGraphContext v0.1.0 · Documentation

Database Backends

CodeGraphContext is backend-agnostic. It supports multiple graph database engines to balance performance, ease of use, and scalability.

Comparison Overview

Backend Mode Best For Complexity
KùzuDB Embedded Local development, quick starts Zero Config
FalkorDB Embedded/Remote High-performance semantic queries Low to Medium
Neo4j Server Enterprise-scale, visualization High

1. KùzuDB (Embedded)

KùzuDB is the default recommendation for most users. It is an extremely lightweight, in-process graph database.

  • Zero Infrastructure: No Docker or separate services required.
  • Performance: Optimized for analytical queries (OLAP) on graph data.
  • Storage: Saves the graph as a directory on your local machine (usually in .codegraphcontext/).

Installation: 

pip install real_ladybug

2. FalkorDB (High Performance)

FalkorDB is a low-latency graph database designed for massive throughput.

  • Speed: One of the fastest graph engines available for relationship traversal.
  • Modes:
    • FalkorDB Lite: Runs as an embedded engine (Linux/macOS only).
    • FalkorDB Remote: Connects to a standard Redis-based FalkorDB instance.

Installation (Lite): 

pip install falkordblite

3. Neo4j (Enterprise)

Neo4j is the industry standard for graph databases. Use Neo4j when you need advanced visualization or are working with exceptionally large datasets.

  • Visualization: Use the Neo4j Browser (localhost:7474) to see your code graph in a rich, interactive UI.
  • Scalability: Handles billions of nodes and relationships across distributed environments.

Setup: 1. Start Neo4j via Docker or use AuraDB. 2. Configure CGC to use Neo4j: 

cgc config set-db neo4j
cgc config set-neo4j-uri bolt://localhost:7687

Backend Selection Logic

CGC follows a specific priority when deciding which backend to use:

  1. Explicit: Any database specified via the --database flag or CGC_RUNTIME_DB_TYPE env var.
  2. Configured: The default set via cgc config db.
  3. Automatic:
    • Remote FalkorDB: If FALKORDB_HOST is detected.
    • Unix: Tries FalkorDB Lite → KùzuDB → Neo4j.
    • Windows: Tries KùzuDB → Neo4j.