#DecPost #22: # AI Code Reviewer

Last time I was mostly working on the UI and getting the basic structure in place. This time I started working more on the actual review logic.

I added different review modes so the AI doesn’t just give the same generic review every time:

• Full Review
• Bugs
• Security
• Performance
• Code Quality
• Readability
• Tests

Each mode now has its own specific prompt. So if I choose Bugs the AI should focus on finding bugs and incorrect behavior instead of trying to review everything at once.

I also worked on making the selected review type control what gets sent to the AI.

What’s left

• Connect the actual AI API
• Send the code + selected review type
• Display the AI response
• Add loading and error handling
• Test all the review modes
• Improve the prompts based on the results
• Test with different programming languages
• Clean up anything that breaks during testing

What’s next

Next I’m going to work on connecting the actual AI and finally see how the reviewer performs with real code.