graphify-labs/graphify v0.9.91 is a Python tool that turns a codebase and related materials into a queryable knowledge graph. Its inputs include code and documents, SQL schemas, configuration files, and PDFs. Instead of digging through files directly, it is designed so you explore the graph to find the connections between materials. It also provides the /graphify skill used in Claude Code, Cursor, Codex, and Gemini CLI. In other words, not only code but also documents, data structures, configuration, and PDFs are all placed within the scope of a single graph and queried together. This structure — gathering different kinds of materials into one graph and querying the connections between them — is the center of the tool. Materials other than code enter the same graph.
Code is parsed locally and deterministically using Tree-sitter syntax trees. Each edge is marked with whether it is a fact explicitly stated in the source or an inferred relationship. So the connections in the graph can be distinguished by the nature of their evidence. Not using embeddings or a vector store is also a characteristic of the design. It is written in Python and licensed under Apache-2.0. There is also a separate cloud paid service. The targets of the provided skill are not only Claude Code and Cursor but also Codex and Gemini CLI.
However, labeling the type of a connection and verifying the accuracy of inference are different things. The provided information includes no numbers showing the accuracy or speed of graph traversal, or performance by scale. Therefore, it is difficult to judge the actual search quality or performance on a large codebase from this description alone.