Moderne’s ground-breaking system for code refactoring at scale enables developers to know their code like never before and automate change across multiple repositories at once.
The LST is an evolution of the abstract syntax tree, providing the most detailed code representation available today. It captures rich, semantic-based data including all dependencies (transitive and direct), type information, and formatting—everything needed to search and transform code programmatically. Moderne serializes LSTs to disk to horizontally scale change across your repositories.
Recipes are deterministic programs that accurately search and fix your code by navigating to different parts of the LST tree. Recipes can perform simple tasks like finding a method to complex tasks comprising 1,000s of steps like performing major migrations or mass security vulnerability sweeps. Recipes are created and maintained by the open-source OpenRewrite community, and developers can easily build custom recipes.
The Moderne refactoring system can call on AI large language models (LLMs) to support the work being done on your codebase. The LST is a powerful retrieval augmented generation (RAG) engine for making LLMs much smarter at understanding and modifying code, enabling greater efficiency and accuracy. We are able to select the best OSS LLMs for the task at hand.
Within the Moderne Platform, AI models are implemented as sidecars and support AI-powered recipe searches, AI-assisted code transformations, and AI-driven code diagnostics with recommended, actionable fixes.
Moderne is able to scale refactoring tech that is idiomatic to different communities, such as codemods and linters. We integrate them in recipes and run them on code as text within the platform across multiple repositories at once. This enables developers to get a ‘one-stop-shop’ approach to access and execute refactoring tasks at mass scale. Moderne will continue to support languages natively as well.