Code Review Graph

Code Review Graph is an open-source, local-first code intelligence graph with 32,200+ stars on GitHub that helps AI coding tools read only the code that matters.

Overview

AI coding assistants waste enormous context re-reading whole repositories. Code Review Graph builds a persistent, incremental map of your codebase — files, symbols and their relationships, parsed with tree-sitter — and serves it to AI tools over MCP and a CLI. When an assistant reviews a change or explores a large repo, it can query the graph for exactly the relevant slice instead of swallowing thousands of irrelevant lines, and the project publishes benchmarks showing large context reductions on review and large-repo workflows. Everything is local-first: the graph lives on your machine, updates incrementally as code changes, and installs from PyPI. It works alongside tools like Claude Code and other MCP-capable agents. The project is MIT licensed and actively maintained, updated in October 2026.

The economics are the pitch: frontier-model context is the most expensive input in AI-assisted development, and most of it is wasted on files irrelevant to the task at hand. By front-loading the indexing work locally and incrementally, Code Review Graph turns every subsequent agent session into a cheaper, faster one — reviews focus on the changed code and its true dependents, and questions about a large repo stop requiring a full-repo scan. It is a pragmatic, unglamorous optimization that compounds daily, which is exactly why it spread so quickly through the MCP ecosystem.

Key Features

• Persistent code graph – a durable map of files, symbols and relationships across your repo.
• MCP and CLI access – AI coding tools query the graph through the Model Context Protocol.
• Incremental updates – tree-sitter parsing keeps the graph fresh as code changes.
• Context reduction – benchmarked savings on reviews and large-repository workflows.
• Local-first – your code and graph stay on your machine; nothing is uploaded.
• Easy install – available on PyPI as code-review-graph.

💳 Pricing

Code Review Graph is completely free and open source under the MIT license.

Who Is It For

Developers using AI coding agents on medium-to-large codebases who want faster, cheaper, more focused reviews and code exploration.