GitLab Knowledge Graph (gitlab-org.gitlab.io)

GitLab has launched the Knowledge Graph in public beta — a code analysis engine that builds a queryable, structured map of your repositories and links it to the GitLab SDLC. The system parses code to extract structural elements (files, directories, classes, functions, modules) and code relationships (function calls, inheritance hierarchies, module dependencies), and persists that graph in a high-performance graph database called Kuzu. The result is a scalable, explorable knowledge base you can query to surface precise code context across large codebases. For the AI/ML community this is significant because it provides a ready-made retrieval layer and rich semantic signals for RAG pipelines, automated code reasoning, and model-assisted development. Practical uses include feeding vector/RAG systems with exact code-context hits, generating architectural visualizations and dependency diagrams, and powering advanced code-intelligence features (impact analysis, call‑graph queries, cross-repo search). The project is in public beta and meant for evaluation and feedback; an installation guide and an open “First Iteration” epic are available for contributors.