Graphify Transforms Projects into Queryable Knowledge Graphs, Streamlining Code and Research Analysis

Graphify is gaining significant traction within the open-source community for its unique capability to transform diverse project files—including code, documents, videos, and images—into a connected knowledge graph. Inspired by Andrew Karpathy’s ‘second brain’ concept, Graphify aims to represent the relationships between disparate pieces of information, facilitating more efficient understanding and querying. This approach positions Graphify as a simpler, more maintainable alternative to traditional Retrieval-Augmented Generation (RAG) systems for contextual information retrieval, particularly beneficial for large-scale projects and comprehensive research endeavors.

The implementation of Graphify is straightforward, requiring Python 3.10+ and the UV virtual environment manager. Once installed, an AI agent, such as Claude Code, scans the project to generate the initial graph. This process, while token-intensive initially, results in an interactive HTML graph, a summary report, and a JSON file detailing data relationships. Graphify’s core utility lies in its query capabilities, allowing users to leverage commands like grapify query for specific questions and grapify path to trace connections between concepts or files. This graph-based querying significantly reduces token consumption in subsequent searches and enhances the accuracy and contextual richness of responses, effectively functioning as a ‘personal wiki’ for codebases or research data. Furthermore, Graphify integrates seamlessly with tools like Obsidian vaults and graph databases such as Neo4j, offering flexible options for persistent graph storage and interaction.