Static analysis · 5 tools
Top 5 static analysis tools for detecting code smells, unused code and duplication in Java
PMD, SonarQube and Error Prone are the first three tools on this shortlist, which is written for Java maintainers who are refactoring large codebases and need to find code smells, unused code and copy-paste. PMD comes first because its documentation covers unused variables, empty catch blocks and a built-in copy-paste detector. The remaining tools cover platform-level quality gates, compile-time checking, Kotlin smell analysis and technical-debt prioritization.
| # | Tool | Best for | Key feature | Platforms / deployment | Pricing | More |
|---|---|---|---|---|---|---|
| 1 | PMDPMD | Java teams that want rule-based smell checks and duplicate detection in one open tool | More than 400 built-in rules, extendable with custom rules written in Java or as XPath queries | Command line (check command); Windows install via Chocolatey; Eclipse plugin and other IDE and CI plugins | See vendor pricing page | Introduction |
| 2 | SonarQubeSonar | Teams that want maintainability and technical-debt checks enforced at merge time | Analyzes branches and pull requests for bugs, maintainability issues and technical debt across 45+ languages | Managed cloud (SonarQube Cloud) or self-managed server (SonarQube Server), including air-gapped; common editors and major Git platforms | See vendor pricing page | Introduction |
| 3 | Error ProneGoogle | Java teams that want refactoring mistakes caught during the build | Augments the compiler's type analysis to catch mistakes before they become production bugs | Java build integration, for example Bazel Java compilation | See vendor pricing page | Introduction |
| 4 | detektdetekt | Mixed Java and Kotlin codebases that need smell analysis on the Kotlin side | Code smell analysis on Kotlin projects | Gradle, Maven and Bazel; Android, JVM, JS, Native and Multiplatform projects | See vendor pricing page | Introduction |
| 5 | CodeSceneCodeScene | Engineering teams prioritizing which parts of a large codebase to refactor | Identifies and prioritizes high-impact technical debt using the CodeHealth™ metric | Hosted cloud or on-prem on your own machine or server; connects to GitHub, GitLab, BitBucket and Azure DevOps; IDE and CLI | See vendor pricing page | Official site |
1 PMD
By PMD
PMD is an extensible, cross-language static source code analyzer that finds common programming flaws such as unused variables, empty catch blocks and unnecessary object creation. It focuses mainly on Java and Apex.
Why it is here: It sits first because its documentation names unused variables explicitly and bundles CPD, a copy-paste detector, which covers two of the three needs in this use case in one tool.
- Best for
- Java teams that want rule-based smell checks and duplicate detection in one open tool
- Key features
- More than 400 built-in rules, extendable with custom rules written in Java or as XPath queries
- Finds unused variables, empty catch blocks and unnecessary object creation
- CPD copy-paste detector for duplicated code
- Parses source into abstract syntax trees using JavaCC and Antlr before running rules
- Eclipse plugin released alongside PMD, plus other IDE and CI plugins
- Limits
- The documentation lists unused variables as an example of what PMD finds; it does not describe whole-project dead-code analysis, so check the rule documentation for the unused-code cases you care about. Commercial support such as consulting and training comes from third-party companies.
- Platforms / deployment
- Command line (check command); Windows install via Chocolatey; Eclipse plugin and other IDE and CI plugins
- Pricing
- See vendor pricing page (not quoted here; facts checked 21 November 2026, Asia/Dhaka)
Official website (pmd.github.io) · Read our introduction to PMD
2 SonarQube
By Sonar
SonarQube is a code quality and security platform that automates reviews and verification of human-written and AI-generated code. It is available as a managed cloud service or a self-hosted server.
Why it is here: It follows PMD because it adds branch and pull request analysis for maintainability issues and technical debt, which suits ongoing refactoring work more than one-off scans.
- Best for
- Teams that want maintainability and technical-debt checks enforced at merge time
- Key features
- Analyzes branches and pull requests for bugs, maintainability issues and technical debt across 45+ languages
- Quality gates decide what merges and record every check
- Quality gate results appear in pull requests on major Git platforms
- AI CodeFix suggests fixes, and the Remediation Agent opens verified pull requests
- Architects define the intended structure and deviations are flagged in quality gates
- Limits
- The documented scope is maintainability issues, bugs and technical debt; duplicate-code and unused-code detection are not listed in the facts we checked, so confirm them in the vendor documentation. SCA requires SonarQube Advanced Security.
- Platforms / deployment
- Managed cloud (SonarQube Cloud) or self-managed server (SonarQube Server), including air-gapped; common editors and major Git platforms
- Pricing
- See vendor pricing page (not quoted here; facts checked 21 November 2026, Asia/Dhaka)
Official website (sonarsource.com/products/sonarqube) · Read our introduction to SonarQube
3 Error Prone
By Google
Error Prone is a Google-developed, open-source tool that augments the Java compiler's type analysis to catch mistakes at build time. Google uses it in its own Java build system.
Why it is here: It sits third because it is documented as a compile-time bug checker rather than a smell or duplication detector, but it is useful during refactoring for code that no longer does what was intended.
- Best for
- Java teams that want refactoring mistakes caught during the build
- Key features
- Augments the compiler's type analysis to catch mistakes before they become production bugs
- Hooks into the standard build so all developers run it automatically
- Findings surface as compilation errors, for example in a Bazel Java compilation
- Each finding carries a named bug pattern, such as CollectionIncompatibleType, with a descriptive message
- Produces suggested fixes, so tooling can be built on top of it
- Limits
- Documented as a bug-pattern checker for Java builds; it is not described as a duplicate-code or dead-code detector, so pair it with a tool that covers those. Teams on build systems other than those in the documentation should check the vendor documentation.
- Platforms / deployment
- Java build integration, for example Bazel Java compilation
- Pricing
- See vendor pricing page (not quoted here; facts checked 21 November 2026, Asia/Dhaka)
Official website (errorprone.info) · Read our introduction to Error Prone
4 detekt
By detekt
detekt is an open-source static code analyzer for Kotlin that detects code smells and is configurable and extensible. It is developed by the community.
Why it is here: It sits fourth because its documented code smell analysis targets Kotlin, so it matters mainly for Java projects that also contain Kotlin code.
- Best for
- Mixed Java and Kotlin codebases that need smell analysis on the Kotlin side
- Key features
- Code smell analysis on Kotlin projects
- Baselines let legacy projects suppress existing issues while ensuring no new issues are introduced
- Complexity reports based on lines of code, cyclomatic complexity and number of code smells
- Reports in HTML, Markdown, SARIF, Checkstyle (XML) and custom formats
- First-party Gradle plugin, plus plugins for Maven and Bazel
- Limits
- Documented for Kotlin only; it is not described as analyzing Java source, so Java-only projects should look at the other tools first. Duplicate-code and unused-code detection are not listed in the facts we checked.
- Platforms / deployment
- Gradle, Maven and Bazel; Android, JVM, JS, Native and Multiplatform projects
- Pricing
- See vendor pricing page (not quoted here; facts checked 21 November 2026, Asia/Dhaka)
Official website (detekt.dev) · Read our introduction to detekt
5 CodeScene
By CodeScene
CodeScene is a code quality and technical-debt analysis platform built around the CodeHealth™ metric. It helps teams prioritize debt, gate pull requests and safeguard AI-generated code.
Why it is here: It closes the list because it helps decide where to refactor first and guards against new debt, rather than listing individual smells, unused code or duplicates.
- Best for
- Engineering teams prioritizing which parts of a large codebase to refactor
- Key features
- Identifies and prioritizes high-impact technical debt using the CodeHealth™ metric
- Pull request gates can block merges that drop Code Health below a configured bar
- Real-time CodeHealth™ feedback in the IDE
- CLI and REST API for pipelines and internal systems
- Insight into knowledge distribution and bus-factor risk
- Limits
- The documentation does not say which languages it analyzes or describe duplicate-code or unused-code detection, so Java teams should check the vendor documentation for language coverage.
- Platforms / deployment
- Hosted cloud or on-prem on your own machine or server; connects to GitHub, GitLab, BitBucket and Azure DevOps; IDE and CLI
- Pricing
- See vendor pricing page (not quoted here; facts checked 21 November 2026, Asia/Dhaka)
How we chose these tools
We looked at three things: fit with the use case of finding smells, unused code and duplication in big Java codebases; the capabilities each vendor documents; and the deployment and platform options described in the documentation, such as build-system plugins, IDE support and self-hosting. Tools that document smell and duplication detection directly are placed first, and tools whose documented focus is compile-time checking, a different language or debt prioritization follow.
All facts come from official vendor pages and are checked against them at the time of writing. The order reflects editorial fit for this specific use case and is explained in each entry; it is not a statement that one tool is better than another overall. The list is built from documentation rather than product trials, so where a capability is not documented we say so and point you to the vendor documentation.
Order used in this guide: Ordered by smell and duplication detection strength, with compile-time checkers after.
Frequently asked questions
Which tool finds duplicated code in Java?
PMD includes CPD, a copy-paste detector that finds duplicated code in many languages. The other tools on this list do not document duplicate detection in the facts we checked.
Can PMD find unused code?
PMD's documentation lists unused variables, empty catch blocks and unnecessary object creation among the flaws it finds, using more than 400 built-in rules. Check the rule documentation for other unused-code cases.
Can I run these tools in a pull request workflow?
SonarQube shows quality gate results in pull requests on major Git platforms, and CodeScene's pull request gates can block merges below a configured Code Health bar. Error Prone surfaces findings as compilation errors in the build.
Does detekt work on Java code?
detekt is documented as a static analyzer for Kotlin, with Gradle, Maven and Bazel plugins. It is relevant to Java projects mainly where Kotlin code sits alongside the Java.
How can I adopt a smell checker on a legacy codebase without a wall of warnings?
detekt documents baselines that suppress existing issues while ensuring no new issues are introduced, though it applies to Kotlin. For Java, SonarQube's quality gates and CodeScene's pull request gates are documented ways to control what new code can merge.
- PMD: pmd.github.io — checked 21 November 2026
- SonarQube: sonarsource.com/products/sonarqube — checked 21 November 2026
- Error Prone: errorprone.info — checked 21 November 2026
- detekt: detekt.dev, detekt.dev/docs/intro — checked 21 November 2026
- CodeScene: codescene.com, codescene.com/product/free-trial, codescene.com/product/code-health-mcp — checked 21 November 2026
Linthelm is independent and not affiliated with the vendors listed. No vendor paid for inclusion or position. Product names belong to their owners. Features and prices change; confirm details on each official website.