code-review-graph

Code Review Graph (CRG) revolutionizes code analysis by building a persistent, local graph of your codebase. This structural memory layer provides efficient context to various code assistants, enabling faster, sharper, and more cost-effective code reviews. Key capabilities of CRG include: * Persistent local code graph * 8.2x average context reduction * Over 30 MCP tools and 5 ready-made prompts * Local-first processing, ensuring code never leaves your machine * Incremental re-parsing in milliseconds CRG addresses the "context problem" by preventing assistants from re-reading entire repositories with every prompt. Instead, it delivers only the relevant call chains, code owners, and risks associated with a specific diff. This targeted approach significantly reduces the tokens needed per review, leading to an average 85% reduction in context size and an improved reviewer signal quality of 8.8/10. It processes a five-stage pipeline locally and incrementally, keeping context fresh without rereading the entire codebase. This tool is ideal for engineering teams that frequently use code assistants. CRG ensures that your team benefits from precise, contextual code analysis, keeping your development cycle lean and efficient while maintaining high code quality. It works with popular tools like Claude Code, Cursor, Copilot, Codex, Windsurf, and Zed through MCP compatibility.