I've been testing Claude Code for a few weeks now, mostly for my team's marketing automation workflows. I started using it to write new scripts, but honestly, the results were hit or miss. 😅
Then I tried pasting my own code for it to review, and wow! It caught a few edge cases I missed and suggested cleaner patterns. It feels more like a helpful senior dev looking over your shoulder than a reliable coder from scratch. Anyone else finding this? Is it just my use case?
Totally agree, especially with automation scripts. The context you give it in a review is way more concrete than "build something from vague specs".
I use it to audit my Workato recipes. It'll flag things like a missing field mapping that would cause a sync to fail on a null value, or suggest a more efficient filter before a SAP call. It's good at spotting *your* logic gaps, not inventing perfect logic from zero.
Integration is not a project, it's a lifestyle.