Local, high level, semantic code reviews (not only syntax)

Ideas Feature Requests
1

Feature request for product/service

Cursor IDE

Describe the request

Hello, Is there a way to trigger a code review with high level semantic analysis in the Cursor IDE?

I am testing a prompt for doing local, self code reviews

But it’s not that easy to visualize the changes

I want to be able to control the model used for the review and do high level reviews not only syntax based.

Sometimes the system does an error which is semantic but harder to find, and requires me to take a look at the code review, would be good to do this process faster with the help of AI locally or to be able to trigger a background agent to find these and fix them.

Operating System (if it applies)

Linux

2

Hey! In Cursor, there are a few ways to do AI code review:

  1. Agent Review (built-in feature):
  • In the Source Control tab, you can run Agent Review to analyze all local changes
  • After the agent finishes, click Review > Find Issues to analyze a specific diff
  • Docs: Review | Cursor Docs
  1. Cursor CLI for custom reviews:
    You can use the Cursor CLI locally with your own prompts for high-level semantic analysis:
cursor-agent --model "claude-sonnet-3.5" --print "Review this codebase for semantic issues, architectural problems, and logic errors. Focus on high-level analysis, not just syntax."
  1. Bugbot for automatic PR reviews:
    Bugbot | Cursor Docs

For your use case, I’d recommend trying the Cursor CLI since it gives you full control over the model and prompts for semantic analysis.