The Six-Phase Indexing Pipeline

How GitNexus transforms raw source files into a queryable knowledge graph

Source Files PHASE 1 Structure File tree walking Folder/file PHASE 2 Parsing Tree-sitter AST extraction 13 languages PHASE 3 Resolution Cross-file linking with confidence PHASE 4 Clustering Leiden community detection PHASE 5 Processes Execution flow tracing Entry points PHASE 6 Search BM25 + semantic + RRF fusion Knowledge Graph Nodes + Edges Clusters + Processes Search indexes 7 MCP Tools AI agents File walker Tree-sitter Language-aware Graphology LadybugDB
Interactive Diagram: Hover over any phase to see details, or click "Step Through Pipeline" to animate the indexing flow.