Introducing Code Review, a new feature for Claude Code. When a PR opens, Claude dispatches a team of agents to hunt for bugs.

Claude reviewing Claude?

Agents search for bugs in parallel, verify each bug to reduce false positives, and rank bugs by severity. You get one high-signal summary comment plus inline flags.

We've been running this on most PRs at Anthropic. Results after months of testing: PRs w/ substantive review comments went from 16% → 54% <1% of review findings are marked incorrect by engineers On large PRs (1,000+ lines), 84% surface findings, avg 7.5 issues each

Code Review optimizes for depth and may be more expensive than other solutions, like our open source GitHub Action. Reviews generally average $15–25, billed on token usage, and they scale based on PR complexity.

Code Review is available now as a research preview in beta for Team and Enterprise. Read the blog for more:

Very soon the only thing we require on the keyboard is Claude key.

Babe wake up , new Claude update just dropped

Where do you see yourself in 5 years? Software engineer:

I didn’t expect Claude to talk to me in this way😭

Claude writes code Claude reviews code Claude finds bugs written by Claude Claude suggests a fix to Claude

cheers, i think this is a good feature. well done

How can I thank you for making my job easier by just taking it away.

This is great, but you need to be careful not to treat this as a complete solution for verifying your code. The system that proposes a solution shouldn't be the one trying to find its flaws. Same model, same training biases, same blind spots.

Charge customers to write AI slop and charge again to review the AI slop 📈

😂😂

I love my code reviewers.

this is ridiculous.

You are at *Claude*

Claude reviewing its own code

ANOTHER DAY ANOTHER CLAUDE UPDATE

How many startups are doomed this time? 👀

Software engineers watching Claude automate their job in real time

Hmm... Using the same AI to write the code and review the code feels like grading your own homework.

Claude to software engineers

So now Claude will review Claude and alert Claude about the mistake of Claude

If Claude writes the code and reviews it, isn’t that just grading its own homework? Code generation and code verification are different engineering problems. Independent review layers exist for a reason.

Nice

@lydiahallie Wait, why does Claude Code need code review when it can just write bug-free code in the first place? Or was this made for people that code with something else?

Why doesn't Claude write fewer bugs to begin with?

My guys can you just stop shipping? I’m still catching up from last week

Wouldn't you want the system reviewing your code to be architecturally independent from the one that wrote it? Same model, same training, same optimization goals = same blind spots. I'm not sure if that sounds like a review. Instead, a second read.