What is Code Review Graph?

The GitHub repository code-review-graph is an open-source Python tool that helps AI coding assistants understand your codebase more efficiently. Instead of making an AI read thousands of files every time you ask a question, it builds a persistent knowledge graph of your project so the AI reads only the files that matter. This can dramatically reduce token usage and speed up code reviews, debugging, and architecture analysis.

• Without Code Review Graph: AI scans hundreds or thousands of files to answer a question.
• With Code Review Graph: AI already knows the relationships between files and reads only the relevant ones.

1. Smart Code Reviews

Instead of reviewing every changed file, it identifies the blast radius—the functions, files, and tests likely affected by a change.

payment.service.ts changed

↓

OrderService
InvoiceService
PaymentController
WebhookHandler
Payment Tests

This helps reviewers focus on the parts of the project that actually matter.

2. Huge Token Savings

One of its biggest goals is reducing AI token usage.

According to the project's benchmarks, graph-based queries can reduce context from hundreds of thousands of tokens to just a few thousand, depending on the repository and task.

3. Incremental Updates

After the first build, it doesn't re-index the whole project.

Instead, it updates only the files that changed, making refreshes much faster.

4. Semantic Search

Search using natural language, for example:

• "How does authentication work?"
• "Where is JWT implemented?"
• "Find payment logic."

The graph helps the AI locate the most relevant code instead of performing simple text searches.

5. Architecture Analysis:

It can automatically identify:

• Core modules
• Hub nodes
• Bridge nodes
• Connected components
• Community structure

This makes it easier to understand unfamiliar or very large projects.

6. Interactive Visualization:

The tool can generate an interactive graph showing how files and modules are connected.

Useful for:

• Learning a new project
• Explaining architecture
• Refactoring safely

7. Wiki Generation:

 It can generate Markdown documentation from your project's structure, helping teams maintain internal documentation.

Supported Languages

The repository supports many languages, including:

• Python
• JavaScript
• TypeScript
• React
• Go
• Rust
• Java
• C/C++
• C#
• PHP
• Kotlin
• Swift
• Dart
• SQL
• Vue
• Svelte
• Astro
• Jupyter Notebooks

Supported AI Tools

It integrates with several AI coding assistants, including:

• Codex
• Claude Code
• Cursor
• GitHub Copilot
• Gemini CLI
• Continue
• Kiro
• Qwen
• OpenCode

and other MCP-compatible tools.

Install:

pip install code-review-graph

Configure your AI tool:

code-review-graph install

Build the graph:

code-review-graph build

Check project status:

code-review-graph status

Watch for changes automatically:

code-review-graph watch