Graphify: Turn Codebases into Knowledge Graphs to Slash AI Token Costs

Graphify-Labs has released graphify, an open-source tool that converts entire codebases into structured knowledge graphs to reduce AI token consumption by up to 71.5x. The tool uses tree-sitter for local static analysis and an AI semantic pass on documentation, providing agents with instant architectural awareness and integrating with CLI agents and IDEs.

AI coding agents such as Claude Code, Cursor, and GitHub Copilot excel at generating code snippets, but they often struggle to maintain full architectural context in large software repositories. When answering multi-file questions, traditional agents repeatedly search and re-read raw source files—wasting thousands of context window tokens and introducing latency. graphify is an open-source tool developed by Graphify-Labs to solve codebase context fragmentation. By converting entire repositories into structured, queryable knowledge graphs, graphify provides AI agents with instant architectural awareness while reducing token consumption by up to 71.5× . graphify bridges the gap between deterministic static analysis and semantic AI comprehension. It acts as an architectural layer that maps relationships between functions, classes, modules, and documentation across your entire project. The structural analysis phase of graphify uses tree-sitter to parse your code into an Abstract Syntax Tree AST . It extracts exact function calls, class inheritance, and import dependencies locally—requiring zero LLM tokens and sending zero data to external servers. Beyond static code structure, graphify uses your AI assistant to perform a semantic pass on documentation, Markdown files, and architectural notes. It links high-level design concepts directly to the underlying source code nodes. By providing pre-computed graph queries rather than forcing the LLM to grep through hundreds of raw files, graphify dramatically reduces prompt token volume, enabling faster responses and lower API bills. Running /graphify generates multiple architectural artifacts inside your workspace: graph.html : An interactive D3 visual map of your repository. GRAPH REPORT.md : A structured markdown summary of codebase architecture. graph.json : A structured graph format compatible with GraphRAG pipelines. graphify integrates with popular CLI agents and IDEs, and can be configured as a Git pre-commit hook to keep your knowledge graph continuously synchronized. Install Graphify uv tool install graphify-labs Generate codebase knowledge graph graphify By shifting AI codebase analysis from brute-force file reading to structured knowledge graph querying, graphify sets a new standard for developer productivity. It provides AI agents with the deep architectural awareness required for large-scale refactoring and system design. Want to map your codebase? Check out the graphify GitHub Repository.