CodART
github.com
- Free plan
- Linux
- Web
- API
- Self-hosted
What CodART is
CodART is a toolkit for automated refactoring and multi-objective optimization of Java source code. It documents 47 refactoring operations, including Extract Class, Move Method, and Encapsulate Field, and supports project-level and batch refactoring. Its optimization uses NSGA-II and NSGA-III across eight QMOOD quality metrics. The toolkit also includes PPO-based refactoring sequence generation and models that predict testability from source code metrics. For code quality checks, it integrates PMD with custom rulesets for detecting code smells. CodART uses SciTools Understand for code parsing and metrics computation. Its documented interfaces are a React web interface and a FastAPI REST API, and Docker Compose instructions cover running services locally. The project is available under an MIT License, whose stated terms permit use, copying, modification, distribution, sublicensing, and sale subject to its conditions. However, the documented setup requires a separate SciTools Understand professional license; academic licenses are available for research. The quick-start specifies at least 8 GB RAM and 4 CPU cores, with 12 GB or more recommended for large projects. The project welcomes contributions and questions by email.
Who it suits
CodART suits developers and researchers working with Java who need automated refactoring, optimization, or code quality analysis. Its documented setup has separate licensing and system requirements to consider.
Good to know, and what to watch
- Documents 47 automated refactoring operations.
- Supports project-level and batch refactoring.
- Includes optimization across eight QMOOD metrics.
- Provides a web interface and REST API.
- MIT License permits use and modification under its conditions.
- Requires a separate SciTools Understand professional license.
- Quick-start requires at least 8 GB RAM and 4 CPU cores.
- Large projects are recommended to have 12 GB or more.
Plans
Every plan and priceMIT licensed software
Open source
Getting started
- Clone or open the CodART GitHub repository
- Review the MIT License terms
- Obtain a SciTools Understand professional license, or an academic license for research
- Prepare a system with at least 8 GB RAM and four CPU cores
- Use the Docker Compose instructions to run services locally
- Access the React web interface or FastAPI REST API
Limits to check
CodART is free under the MIT License, but the documented setup depends on a separate SciTools Understand professional license. The quick-start specifies at least 8 GB RAM and four CPU cores, with 12 GB or more recommended for large projects.
The Club's review
CodART is a Java toolkit for automated, project-wide refactoring, optimization, and code-quality analysis. It is best suited to engineering teams and researchers working across substantial codebases, especially those interested in optimization or experimentation rather than one-off editor fixes. The main trade-off is that its documented analysis workflow depends on a separate SciTools Understand professional license and a reasonably capable machine.
Overview
CodART combines broad source transformations with quality-oriented optimization and machine-learning components. Its project and batch scope is useful when a change must be coordinated across files and classes. The MIT license allows commercial or research use, modification, redistribution, sublicensing, and sale under its conditions, so organizations can adapt the toolkit without a software purchase. Java is the supported language.
Key features
CodART provides 47 automated refactoring operations, including Extract Class, Move Method, and Encapsulate Field. Project-wide and batch execution make this more suitable for systematic modernization than tools limited to local IDE actions. Teams that already know the transformation they want may find the breadth valuable; teams seeking only a small set of predictable edits may not need the full toolkit.
For multi-objective optimization, the project uses NSGA-II and NSGA-III across eight QMOOD quality metrics. That gives researchers and architects a way to explore competing quality goals instead of optimizing a single measure. PPO-based generation of refactoring sequences and testability-prediction models built from source-code metrics add an AI-assisted research angle, although they introduce complexity beyond a conventional refactoring workflow.
PMD integration, including custom rulesets, supports code-smell detection. SciTools Understand performs parsing and metrics computation, tying the analysis pipeline to that external product. A React web interface offers browser access, while a FastAPI REST API supports scripted or service-based integration. Docker Compose instructions provide a documented route to running the services locally.
Pricing
The MIT-licensed software costs nothing. This is a genuinely open-source plan with no paid CodART tier, and the MIT terms permit the broad reuse described above. It fits teams, consultants, and researchers that can manage their own deployment and dependencies.
The important qualification is outside CodART's price: the documented setup requires a separate SciTools Understand professional license. Academic licenses are available for research, but the Understand license is an additional procurement and licensing consideration rather than part of the free CodART plan. CodART's published system guidance calls for at least 8 GB of RAM and four CPU cores, with 12 GB or more recommended for large projects, so infrastructure can also affect the real cost of adoption.
Platforms
CodART is available as an API, on Linux, as self-hosted software, and through the web interface. Local Docker Compose deployment favors organizations that want control over execution and source-code handling. The API and web front end also let teams divide interactive review from automated pipelines.
Who it's for
Choose CodART when Java maintenance spans an entire project and you want refactoring, quality metrics, smell detection, and optimization in one workflow. It is particularly relevant to research groups investigating automated refactoring sequences or testability, and to engineering teams prepared to operate self-hosted services.
It is a weaker fit for developers who want lightweight IDE assistance, support for languages other than Java, or a workflow without a separate commercial analysis dependency. Organizations unwilling to provision the stated CPU and memory should also look elsewhere.
Pros and cons
Pros
- 47 automated operations: covers substantial transformations such as Extract Class, Move Method, and Encapsulate Field.
- Project and batch scope: supports coordinated changes across a codebase rather than isolated edits.
- Optimization and AI features: NSGA-II/III, PPO sequence generation, and testability prediction support experimentation with competing quality goals.
- Open MIT licensing: permits commercial adaptation, redistribution, and integration without a CodART fee.
- API and web access: FastAPI and React provide both programmable and browser-based workflows.
Cons
- Separate Understand license: the documented parsing and metrics workflow depends on an additional professional product.
- Resource requirements: at least 8 GB RAM and four CPU cores are required, with more recommended for large projects.
- Java-only scope: teams maintaining other primary languages cannot use CodART as a general refactoring platform.
- High workflow complexity: optimization and machine-learning capabilities may be unnecessary for straightforward, predefined refactoring.
Alternatives
Code Refactoring Tools is the broader category to browse when you want to compare several approaches.
Pick ReSharper instead for IDE-oriented .NET and C# work, including a free non-commercial plan and a paid plan at $389 a year. Choose Sourcery when you want freemium API, extension, self-hosted, or web tooling and an open-source plan with AI reviews, PR summaries, diagrams, one-click fixes, and custom rules.
OpenRewrite is the simpler choice for Apache-licensed recipes and repository-at-a-time execution on Linux or self-hosted installations. Choose ast-grep for syntax-aware searching and transformation across its supported extension, desktop, API, and web environments.
CodeRush suits users wanting a free commercial-use extension with the same features and updates across its subscriptions. Roslynator is a free option for Roslyn-based tools and extensions. Spoon offers a free, self-hosted Java analysis option without stated usage limits, while Coccinelle is a free Linux and macOS alternative.
Verdict
CodART is a strong choice for Java teams and researchers that need project-wide, batch refactoring combined with quality optimization, testability modeling, and smell analysis under an MIT license. Look elsewhere if you need multi-language or lightweight IDE refactoring, or if obtaining SciTools Understand and meeting the hardware guidance would outweigh the benefits of its unusually broad automation stack.
The verdict
CodART combines Java refactoring operations with optimization, testability prediction, and code smell detection. Its separate SciTools Understand license requirement and stated hardware needs are key setup considerations.
Questions people ask
Is CodART free?
Yes. CodART is free and open source under an MIT License.
What language does CodART support?
CodART supports Java source code.
Does CodART require another license?
Yes. The documented setup requires a separate SciTools Understand professional license; academic licenses are available for research.
How can CodART be run?
The project documents API, Linux, self-hosted, and web environments, with Docker Compose instructions for running services locally.
What interfaces are available?
CodART documents a React web interface and a FastAPI REST API.
What hardware is required?
Quick-start requirements specify at least 8 GB of RAM and four CPU cores. At least 12 GB is recommended for large projects.
What the maker publishes
- Analysis integration
- CodART integrates SciTools Understand for code parsing and metrics computation. source
- Code quality
- The repository describes PMD integration with custom rulesets for code smell detection. source
- Deployment
- The repository provides Docker Compose instructions for running its services locally. source
- External license requirement
- The documented setup requires a SciTools Understand professional license, with academic licenses available for research purposes. source
- Interface and API
- The project documents a React web interface and a FastAPI REST API. source
- License
- The repository includes an MIT License that permits use, copying, modification, distribution, sublicensing, and sale subject to its stated conditions. source
- Machine learning
- CodART includes PPO-based refactoring sequence generation and testability prediction models using source code metrics. source
- Optimization
- The repository describes NSGA-II and NSGA-III optimization of eight QMOOD quality metrics. source
- Purpose
- CodART is a toolkit for automated refactoring and multi-objective optimization of Java source code. source
- Refactoring operations
- The documentation lists 47 automated refactoring operations, including Extract Class, Move Method, and Encapsulate Field. source
- Support
- The repository invites questions by email at [email protected] and welcomes project contributions and discussion comments. source
- System requirements
- The quick-start documentation specifies at least 8 GB RAM and 4 CPU cores, with 12 GB or more recommended for large projects. source
Checked against the maker's pages on 8 Oct 2026.