AI Briefinggraphify: Graph-based exploration that understands code structure without LLMs
Graphify-Labs/graphify
About the project
In an AI coding assistant, entering the /graphify command converts the entire project into a knowledge graph. It connects code, documentation, PDFs, images, and videos into a single graph, enabling structural exploration instead of file search.
This is a pure graph approach that does not use embeddings or vector stores. It leverages tree-sitter ASTs to deterministically analyze function calls, imports, and inheritance relationships across approximately 40 languages, clearly distinguishing between connections explicitly stated in the source and those inferred.
It achieves a recall@10 of 0.497 on the LOCOMO benchmark, demonstrating higher accuracy compared to existing memory systems. Code analysis is processed locally, eliminating data leakage risks, and allows visualization and querying of dependencies in complex codebases without LLM invocation costs.
Graphify-Labs/graphify
Turn any codebase, with its docs, SQL schemas, configs, and PDFs, into a queryable knowledge graph. A /graphify skill for Claude Code, Cursor, Codex, and Gemini CLI: local deterministic AST parsing, every edge explained, no vector store.
Python
This introduction was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report errors, attribution issues, or removal requests via Contact.