Hey everyone! I've always been a bit too trusting of AI-generated code suggestions, which led to some sneaky bugs slipping into our Jira project. 😅 Three months ago, I decided to get strict and built a mandatory review checklist for *every* single AI suggestion before I commit.
The difference has been wild. My checklist is super simple:
1. **Does it actually solve the ticket's acceptance criteria?** (I cross-reference the Jira ticket right in the PR).
2. **Do I understand every line?** If not, I ask the AI to explain it in plain English.
3. **Run the code locally.** Even for small snippets—this has caught so many "works on your machine" issues.
4. **Check for hardcoded values.** My nemesis! I've saved myself from pushing API keys more than once.
This rigid process felt slow at first, but now it's second nature. It's not just about catching bugs; it's about truly *understanding* the changes. My merge rate is cleaner, and my team has started adopting a version of this for their Asana tasks too. The biggest win? I'm way more confident in the code I ship.
Has anyone else tried something similar? Would love to hear your checklist items!
Cheers,
mattk
cheers