Ask HN: How are you reviewing code with Claude Code?

1 learnedbytes 1 7/13/2025, 10:20:39 PM
I'm looking for solutions to automatically review code changes made by Claude Code before they're finalized. Currently considering setting up a secondary AI model with my coding guidelines that would:

Review each edit/commit from Claude Code Either reject with feedback (which I'd pass back to Claude Code) or approve for repo push Essentially act as a gatekeeper between Claude Code and my actual codebase Before I build this myself, wanted to check if anyone has already tackled this problem or has a better approach. Any existing tools, workflows, or architectures you'd recommend?

Thanks in advance for any insights!

Comments (1)

ashwinsundar ยท 1h ago
The point of code review imo is to have a human-in-the-loop actually read the code and decide if it makes sense, in the broader context of the application. That means the human knows things that a linter, formatter, automated test-suite, or even AI agent doesn't know about.

If a computer is doing code review, then it's not a code review. It's an automated check. Even if it's done by AI