Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

LogicLens: Leveraging Semantic Code Graph to explore Multi Repository large systems

Niko Usai, Dario Montagnini, Kristian Ilianov Iliev, Raffaele Camanzo

TL;DR

LogicLens tackles the problem of understanding large, distributed software by building a persistent semantic graph that unifies structural code relationships with domain-level semantics across multiple repositories. It introduces a GraphRAG-based agent that retrieves subgraphs and uses specialized tools to answer technical and workflow queries spanning multiple projects. The method comprises a three-phase graph construction (Structural Graph, Semantic Graph, Semantic Graph Enhancement) and an interactive agent architecture that leverages domain entities to enable cross-repository reasoning and emergent capabilities such as impact analysis and symptom-based debugging. Evaluation on real-world multi-repo systems shows substantial improvements in accuracy and coherence over a baseline vector-based approach, demonstrating practical value for onboarding, debugging, and system understanding in complex software landscapes.

Abstract

Understanding large software systems is a challenging task, especially when code is distributed across multiple repositories and microservices. Developers often need to reason not only about the structure of the code, but also about its domain logic and runtime behaviors, which are typically implicit and scattered. We introduce LogicLens, a reactive conversational agent that assists developers in exploring complex software systems through a semantic multi-repository graph. This graph is built in a preprocessing step by combining syntactic code analysis, via AST parsing and repository traversal, with semantic enrichment using Large Language Models (LLMs). The resulting graph captures both structural elements, such as files, classes, and functions, as well as functional abstractions like domain entities, operations, and workflows. Once the graph is constructed, LogicLens enables developers to interact with it via natural language, dynamically retrieving relevant subgraphs and answering technical or functional queries. We present the architecture of the system, discuss emergent behaviors, and evaluate its effectiveness on real-world multi-repository scenarios. We demonstrate emergent capabilities including impact analysis and symptom-based debugging that arise naturally from the semantic graph structure.

LogicLens: Leveraging Semantic Code Graph to explore Multi Repository large systems

TL;DR

LogicLens tackles the problem of understanding large, distributed software by building a persistent semantic graph that unifies structural code relationships with domain-level semantics across multiple repositories. It introduces a GraphRAG-based agent that retrieves subgraphs and uses specialized tools to answer technical and workflow queries spanning multiple projects. The method comprises a three-phase graph construction (Structural Graph, Semantic Graph, Semantic Graph Enhancement) and an interactive agent architecture that leverages domain entities to enable cross-repository reasoning and emergent capabilities such as impact analysis and symptom-based debugging. Evaluation on real-world multi-repo systems shows substantial improvements in accuracy and coherence over a baseline vector-based approach, demonstrating practical value for onboarding, debugging, and system understanding in complex software landscapes.

Abstract

Understanding large software systems is a challenging task, especially when code is distributed across multiple repositories and microservices. Developers often need to reason not only about the structure of the code, but also about its domain logic and runtime behaviors, which are typically implicit and scattered. We introduce LogicLens, a reactive conversational agent that assists developers in exploring complex software systems through a semantic multi-repository graph. This graph is built in a preprocessing step by combining syntactic code analysis, via AST parsing and repository traversal, with semantic enrichment using Large Language Models (LLMs). The resulting graph captures both structural elements, such as files, classes, and functions, as well as functional abstractions like domain entities, operations, and workflows. Once the graph is constructed, LogicLens enables developers to interact with it via natural language, dynamically retrieving relevant subgraphs and answering technical or functional queries. We present the architecture of the system, discuss emergent behaviors, and evaluate its effectiveness on real-world multi-repository scenarios. We demonstrate emergent capabilities including impact analysis and symptom-based debugging that arise naturally from the semantic graph structure.
Paper Structure (23 sections, 2 figures, 2 tables)

This paper contains 23 sections, 2 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Method
  4. Graph Construction
  5. Structural Graph.
  6. Semantic Graph.
  7. Semantic Graph Enhancement.
  8. Agent Architecture
  9. Projects Tool.
  10. Entities Tool.
  11. Codes Tool.
  12. Graph Query Tool.
  13. Source Tool.
  14. Evaluation
  15. Experimental Setup.
  16. ...and 8 more sections

Figures (2)

  • Figure 1: Comparison between the structural and semantic graphs. (a) Structural Graph: captures static relationships among code units across multiple repositories, including dependencies, method calls, and inheritance. (b) Semantic Graph: enriches the structural graph with domain entities and functional relationships, connecting code units through the operations they perform on shared entities. For example, a Code node representing OrderProcessor is linked to an Entity node Order, illustrating how multiple services collaborate on the same domain concept.
  • Figure 2: ReAct Agent