Detect duplicate code across languages using AST-based similarity tools.
npx skills add https://github.com/mizchi/similarity --skill check-similarity
Please help me install this Agent Skill. Skill: check-similarity Source: https://github.com/mizchi/similarity/tree/main/.claude/skills/check-similarity Command: npx skills add https://github.com/mizchi/similarity --skill check-similarity
Cross-language codebases often contain duplicate logic dispersed across languages, making maintenance costly and error-prone. This Skill detects and highlights semantic duplicates using AST-based analysis and automatically selects the right tool based on file types in the project.
Provide a project path to scan and I will detect languages and run the appropriate similarity tools.
High-intent search queries and answers about installing and using this skill.
To detect cross-language duplicate code, this Skill automatically identifies languages by file extension and runs AST-based similarity analysis tools like similarity-ts and similarity-py to surface exact and near-duplicate logic for consolidation.
AST-based similarity analysis parses code into abstract syntax trees to identify structural duplicates rather than just matching text. This approach finds semantic duplicates across languages, providing prioritized remediation guidance to reduce maintenance costs during refactoring.
Yes, you can use this tool in a polyglot codebase. It detects languages present in the target project and automatically selects the appropriate similarity tool, such as similarity-ts for TypeScript, similarity-py for Python, and similarity-rs for Rust.
The best way to start scanning is to provide a project path. The Skill will automatically detect the languages present based on file extensions, execute the appropriate similarity tools for each detected language, and aggregate the results into a structured report.
This duplicate code detection works by comparing syntax trees. It uses AST-based analysis to identify semantic duplicates and near-duplicates across different languages, ensuring that structurally equivalent code blocks are surfaced even if the text differs.
A key limitation is that detection relies on language-specific tools like similarity-py and similarity-ts, so it is constrained by the supported languages in your project. It aggregates results for exact and near-duplicate blocks, but complex semantic logic translations may not always match perfectly.
jmaglyDecompose large source files into smaller modules with dependency mapping.
kk20300113-pngIdentify shallow, tightly coupled modules and propose deepening refactors.
RelicWare-CoAutomate linting and formatting setup for JavaScript and TypeScript projects.
tao3kSearch code repositories by structure, semantics, or exact text.
