jscpd

jscpd is a copy/paste detector for source code that finds duplicated blocks in more than 225 programming languages and document formats, using the Rabin-Karp algorithm. Copy/paste is a common source of technical debt, and jscpd reports where the duplicated blocks are. The README says the latest versions add a Rust-powered engine.

The newer part of the project is aimed at AI workflows. An `ai` reporter prints compact, token-efficient output (common-path-prefix compression, no code fragments) that can be piped into an LLM prompt; on the project's test fixtures it used about 1,100 estimated tokens against about 5,400 for the default console reporter, roughly 79% fewer. Two agent skills ship with it, a tool-reference skill and a dry-refactoring skill that guides reading clone output and extracting duplicates, installed with `npx skills add kucherenko/jscpd`. A separate jscpd-server package exposes detection as an MCP server over streamable HTTP. The main package provides a CLI and an API, other packages cover the core algorithm, tokenizer, finder, a LevelDB store for large repositories and HTML reporters, and the code is MIT licensed.