Sequencing every public open source repo into a compiler-accurate Lossless Semantic Tree — so the code the world runs on can be understood and improved, at scale.
Enterprise software consists of up to 90% third-party and open source components, stitched together in different ways for business functionality. This software is the basis of the Code Genome Project.
The goal: understand codebases broadly, across open source and enterprise, and share deterministic recipes that advance them. It's all open to developers and their agents alike through code search and a remote MCP server.
Each repository is compiled into a Lossless Semantic Tree: the types, methods, and relationships that define how code behaves. Compiler-accurate, lossless, polyglot*. Each tree holds every signal the compiler sees in a single repository; the project lines them up side by side, so the entire open source ecosystem reads as one corpus. It's the substrate everything else is built on.
Explore the sequenced code of thousands of open source repositories. Search the way a compiler sees code — by type, symbol, and call site, not just strings. Find how any API, annotation, or pattern appears across thousands of repositories in milliseconds. Powered by Moderne's Trigrep search, the code genome is open to explore.
Search the corpus →OpenRewrite recipes rewrite the genome: 10,000+ deterministic programs that transform code precisely and repeatably — whether you're migrating a framework, patching a CVE, or enforcing a standard, the same way every time. Recipes are stored in the project and built to extend, so developers and their agents can adapt a recipe or write a new one.
Browse recipes →A remote MCP server opens the Code Genome to coding agents. They can find any method or type, see how it's really used, and compare versions — all resolved through the type system, not text. Connect in one click.
Connect your agent →* Today the Code Genome Project sequences the Java ecosystem; it's expanding to the full breadth of languages the LST supports, including Python, C#, JavaScript/TypeScript, Scala, and more.
Every OpenRewrite and Moderne recipe, with the artifact you add to get it and its license. Add the artifact to your build to run it — the Maven repository serves them, free for open source.
No recipes match those filters.
A remote MCP server over the entire Code Genome corpus, growing to encompass all the world's provably used open source. Your agent navigates and reads all of open source as if it were fully resolved in an IDE it could click around. Try any tool below, then connect it in one line.
claude mcp add --transport http codegenome https://api.codegenomeproject.org/mcp
Remote MCP over Streamable HTTP at https://api.codegenomeproject.org/mcp. Windsurf, Cline, and Zed take the same URL; stdio-only clients bridge with npx mcp-remote https://api.codegenomeproject.org/mcp.
The exact tools your agent gets, with the exact descriptions it reads — run any of them against the public corpus and read the JSON that comes back. The find_* and usage_examples tools resolve through the corpus's type tables, so they catch what a grep or a docs pipe can't.
A read-only, Maven-compatible repository serving the OpenRewrite OSS recipes, Moderne's licensed recipe modules, and the Moderne CLI — free for open source. Point your build at it and resolve any recipe artifact straight from the catalog.
Declare the Code Genome repository, then add any recipe module as a dependency — versions come from the catalog.
<repository> <id>codegenome</id> <url>https://repo.codegenomeproject.org/maven</url> </repository>
The Moderne CLI runs recipes locally and installs any recipe artifact on demand.
mod config recipes jar install org.openrewrite.recipe:rewrite-migrate-java:LATEST
Free for open source. Apache-licensed recipes run anywhere; MSAL and proprietary modules require a Moderne license.