Code Review Graph is an open-source tool that builds a structural graph of a software repository so AI coding assistants can review changes with more targeted context. It parses code with Tree-sitter, maps symbols and relationships, and uses the graph to identify the files, callers, dependencies, and tests most likely to be affected by a change.

The project exposes this information to AI tools through MCP and can automatically configure supported coding environments. Incremental indexing updates the graph as files change, allowing the system to calculate a change's blast radius without repeatedly asking the model to reread large parts of the repository.

Code Review Graph is useful for developers who use AI for code review, refactoring, or impact analysis on medium and large codebases. Its goal is to reduce unnecessary token consumption while improving review relevance by giving the assistant a machine-generated map of what changed and what those changes may influence.