DEV Community

Cover image for Chat with your MR in Gitlab
Jay Elsheikh
Jay Elsheikh

Posted on • Originally published at thinkreview.dev

Chat with your MR in Gitlab

#gitlab #codereview

Why settle for static code reviews when you can have a conversation?

The Problem We All Know Too Well

You open a GitLab merge request. 437 lines changed across 12 files. You know you should review it thoroughly, but...

// You, staring at the diff:
"Where do I even start? 🤔"
"What changed and why?"
"Are there security issues I'm missing?"
"How do I ask the right questions?"
Enter fullscreen mode Exit fullscreen mode

Sound familiar?

Traditional code reviews are one-directional: you read the code, leave a comment, done. But what if code review was a conversation?

Meet Your New Code Review Partner


ThinkReview is a conversational AI copilot that lives inside your GitLab merge request pages. But here's the game-changer:

🤔 Smart Follow-Up Questions

After analyzing your MR, ThinkReview suggests 3-5 contextually relevant questions you can click to explore:

ThinkReview Smart Follow-Up Questions Demo

Example questions for a typical MR:

🤔 Smart Follow-Up Questions

⬜ Are there any potential security vulnerabilities?
⬜ How will this affect database performance?
⬜ What happens if the API call fails?
⬜ Can you generate a summary comment for the team?
⬜ Are there edge cases we should test?
Enter fullscreen mode Exit fullscreen mode

The magic? These aren't generic. They're generated based on your actual code changes.

Why This Changes Everything

1. Lower Cognitive Load

Before:

You: *stares at 437 lines*
You: "Hmm... what should I check?"
You: *spends 10 minutes thinking*
You: "I'll just approve it"
Enter fullscreen mode Exit fullscreen mode

With ThinkReview:

AI: "Here are 5 things you should explore"
You: *click click click*
You: *actually finds 2 real issues*
Enter fullscreen mode Exit fullscreen mode

2. Junior Developers Learn Faster

The suggested questions become a mentorship tool:

# Junior dev sees:
"What are the performance implications of this code?"

# Junior dev thinks:
"Oh! I should be looking at performance!"

# Junior dev learns:
# Pattern recognition for future reviews
Enter fullscreen mode Exit fullscreen mode

3. More Thorough in Less Time

5 minutes of clicking = what used to take 30 minutes:

  • ✅ Security? Click
  • ✅ Performance? Click
  • ✅ Edge cases? Click
  • ✅ Architecture? Click

Real Example: Catching a Security Issue

The Setup

You're reviewing an authentication feature MR. 186 lines changed.

Step 1: Get AI Review

Click ThinkReview. Within seconds:

## Summary
Implements JWT-based user authentication.

## Key Changes
- Added auth middleware
- JWT token generation/validation
- Protected routes

## Smart Follow-Up Questions
⬜ Are there any potential security vulnerabilities?
⬜ How will this affect API performance?
⬜ What happens if the token expires during a request?
Enter fullscreen mode Exit fullscreen mode

Step 2: Click "Security vulnerabilities?"

ThinkReview responds:

Based on my analysis, I've identified 2 security concerns:

1. **SQL Injection Risk (Line 45)**
Enter fullscreen mode Exit fullscreen mode


javascript
const query = SELECT * FROM users WHERE id = ${userId};

   ❌ User input not sanitized
   ✅ Fix: Use parameterized queries

2. **Missing Auth Check (Line 78)**
Enter fullscreen mode Exit fullscreen mode


javascript
app.post('/api/admin/delete', deleteUser);

   ❌ No permission verification
   ✅ Fix: Add authentication middleware

Would you like me to show you how to fix these?
Enter fullscreen mode Exit fullscreen mode

Step 3: Keep the Conversation Going 

You: "How do I implement parameterized queries in Express?"

AI: *provides specific code examples*

You: "Can you generate a comment for the team?"

AI: *creates formatted GitLab comment*
Enter fullscreen mode Exit fullscreen mode

Generate Professional Team Comments

After your AI conversation, click to generate a summary:

## Code Review Summary

### Overview
JWT authentication implementation with security considerations.

### Key Changes
✅ Auth middleware (auth.js)
✅ Token generation/validation
✅ Protected routes

### Issues Found
1. 🔴 CRITICAL: SQL injection vulnerability (Line 45)
   Recommendation: Use parameterized queries

2. ⚠️ HIGH: Missing authentication check (Line 78)
   Recommendation: Add auth middleware

3. 💡 SUGGESTION: Consider refresh token mechanism

### Verdict
Changes needed before approval.

@developer Please address the security issues above.
Enter fullscreen mode Exit fullscreen mode

Copy. Paste. Done. ✨

The Tech Stack

  • AI Model: Google Gemini 2.5 Pro
  • Context-Aware: Understands language, framework, file relationships
  • Privacy: HTTPS-only, no code storage
  • Works Everywhere: Chrome, Edge, Brave, all Chromium browsers

Common Questions

Q: Won't this make devs lazy?
A: Opposite! It makes reviews more thorough. Suggested questions expose blind spots.

Q: Does it work with self-hosted GitLab?
A: Yes! Works with both gitlab.com and private instances.

Q: How accurate is it?
A: Powered by Gemini 2.5 Pro. Excellent accuracy. But AI complements, not replaces, human judgment.

Try the Challenge

  1. Install ThinkReview (2 min)
  2. Open any GitLab MR
  3. Click 3 suggested questions
  4. Compare to your usual review process

I bet you'll find something you would've missed.

The Bottom Line

Code review shouldn't feel like homework. It should feel like pair programming with an expert who:

✅ Understands your code deeply

✅ Suggests what to look for

✅ Answers instantly

✅ Helps you learn

✅ Makes it enjoyable

Stop reading code. Start conversing with it.

Links

🚀 Install ThinkReview

What's your biggest code review pain point? Drop a comment below 👇

Built by developers who got tired of spending hours on code reviews that should take minutes.

ThinkReview - AI Copilot for GitLab Merge Requests | Free Chrome Extension

Transform your GitLab workflow with AI-powered code reviews. Get instant analysis, security insights, and code suggestions. Free Chrome extension for developers.

favicon thinkreview.dev

Top comments (0)