Early access is in progress. The Chrome extension is coming soon and is not on the Chrome Web Store yet.

One code review. Several AI models. Side by side.

WebCodeReview sends your code or diff to the AI chat sites you are already logged in to, all at the same time, and returns each model's answer. Your AI coding assistant or a script asks; your own browser does the work.

diff.patchauth/session.ts
$

chatgpt

claude

gemini

deepseek

grok

Merged summary (optional)who said what · check against the individual reviews

Example animation: the command, diff and answers are written for this demo and are not real model output.

See it in real tools

How a review looks from inside the AI coding tools you already use.

Claude Code(tested)
Recording comingA screen recording of this tool will appear here. Nothing has been recorded yet.Illustration, not a recording

Why more than one model

Different models catch different problems. In one internal test on a single design document, each of three reviewers raised a serious issue the other two did not. That is one small test, not a benchmark. A single reviewer, human or AI, has blind spots, and a second and third opinion are cheap when they run in parallel.

An illustration of the idea, not data from the test.

How it works

1Ask

Your AI assistant (through MCP) or a script runs wcr ask with the code or diff to review.

2Choose models

Name one or several of ChatGPT, Gemini, Claude, DeepSeek and Grok.

3Your browser does the work

The extension types the prompt into each site in your own logged-in session and reads the answer back.

4Compare

You get each model's answer. One failing site does not stop the others. Optionally, one model merges the reviews into a single report that keeps who-said-what; check the merged report against the individual reviews.

The pieces: a Chrome extension, a small server on your own computer (127.0.0.1), and the wcr command line / MCP server. Read the setup guide.

Join the waitlist

Early access is in progress. The Chrome extension is coming soon and is not on the Chrome Web Store yet.

Questions

Do I need API keys?

No. It uses the AI sites you are already logged in to, with your own accounts and subscriptions.

Where does my code go?

To the AI sites you choose, through your own logged-in browser, and to a server running on your own computer. We do not operate a server that receives your code. Those sites receive the code and will normally keep the conversation in your chat history, so follow your employer's policy on pasting code into third-party AI tools. See the privacy policy.

Is this allowed by the AI sites' terms?

Probably not always. Automating a chat site's web interface may go against that site's terms of service, and the site may limit or suspend accounts it believes are automated. You use your own accounts at your own risk. It acts only on your own account when you ask it to, and does not bypass logins or verification; if a site shows a verification challenge, a person has to solve it.

Will it break when a site changes its layout?

Sometimes, yes. Each site is driven through its web page, so a redesign can break one provider until the extension is updated. The others keep working. All five providers were verified against the live sites on 2026-10-12.

How much will it cost?

Version 1 is free, and code review needs no account. Automated browsing needs you to sign in to WebCodeReview with Google. Paid plans may come later; nothing is decided. See pricing.