Graphify-Labs/graphify Turns Codebases into Knowledge Graphs for AI Assistants
The Python library Graphify-Labs/graphify has trended on GitHub, gaining over 7,900 stars in a single week. It allows developers to convert codebases, documentation, SQL schemas, and configurations into queryable knowledge graphs for AI coding assistants. By bypassing traditional vector stores in favor of deterministic local AST parsing, it provides AI assistants like Claude Code and Cursor with precise, structured context about code relationships. This approach can significantly reduce hallucinations and improve the accuracy of AI-driven code generation and analysis. The tool operates locally without requiring a vector database, explaining every edge in the generated graph. It integrates directly as a skill for popular developer tools including Cursor, Claude Code, Codex, and Gemini CLI.
## BACKGROUND
Traditional Retrieval-Augmented Generation (RAG) for codebases often relies on vector databases, which convert text into mathematical embeddings to find similar content but can struggle with precise structural relationships. In contrast, Abstract Syntax Trees (ASTs) represent the hierarchical structure of source code, and knowledge graphs map explicit entities and their interconnections, enabling deterministic code analysis.