CodeRev

A VS Code extension that uses RAG (Retrieval-Augmented Generation) to perform AI-powered code reviews on GitHub pull requests, grounded in your actual codebase.

How It Works

1. Index your codebase — CodeRev reads your workspace files, chunks them intelligently (Python files are chunked by function/class using AST; other files as whole chunks), embeds them using OpenAI text-embedding-3-small, and stores them in a local ChromaDB vector database.
2. Fetch a PR diff — CodeRev pulls the file changes from any GitHub PR using the GitHub API.
3. RAG review — The diff is embedded and used to retrieve the most relevant chunks from your indexed codebase. The diff + retrieved context is sent to GPT-4o for a grounded, context-aware review.
4. Webview panel — The review is displayed in a VS Code tab.

Features

• Index any codebase with a single command
• Supports TypeScript, JavaScript, Python, Go, Java, C++, C, C#, Ruby, Rust
• Python files are chunked by function and class (AST-based) for higher precision
• ChromaDB persists the index to disk — no re-indexing needed on restart
• Reviews are grounded in your actual codebase, not just the diff in isolation
• Review output displayed in a dedicated VS Code webview panel

Installation

1. Install CodeRev from the VS Code Marketplace
2. Set up the Python backend (required — the extension spawns a local FastAPI server):

   cd <extension-install-path>/backend
   python -m venv venv
   venv\Scripts\activate      # Windows
   pip install fastapi uvicorn openai chromadb python-dotenv
   

3. Create a .env file inside the backend/ folder:

   OPENAI_API_KEY=your_key_here
   

4. Configure settings in VS Code (Ctrl+,, search coderev):

Usage

1. Ctrl+Shift+P → CodeRev: Index Workspace — indexes all supported files in the workspace
2. Ctrl+Shift+P → CodeRev: Review PR — enter a PR number and wait for the review to appear in a new tab

Requirements

• Python 3.10+
• An OpenAI API key
• A GitHub personal access token with repo scope