Always fail to check the contents of terminal

[warengonzaga]

Select Topic Area

Product Feedback

Copilot Feature Area

Copilot Agent Mode

Body

I've been using Agent Mode for Copilot, this is now my default however here's my feedback about it.

1. When doing tests, it always fails to check the contents or the logs for the script it made. If the Agent does not able to see it, it makes multiple tests to confirm if the implementation works. This annoys me as it requires you to make multiple file just to see the actual result of the tests.
2. When I ask to see the past changes I made it will try to look to your git history via git log and git diff which is perfect but there's a downside for git diff. When your git diff is large the bot will hang LOL. It will wait for you to scroll the result and then you have to press Q to end it to get the whole context. It's not that user friendly we should add instruction from the bot what to do with large git diff. Doing that may confuse non-technical users or users that does not have git experience.
3. Always doing an over-engineering step, when I build simple system it tends to build over-engineered enterprise level code even if I explicitly make it simple.
4. Can we make the Agent Mode ask first before implement something? This is the biggest challenge for me as I have to explicitly ask the Agent to confirm first before doing changes. Maybe I miss something here to configure or is this the default behavior?

These are my legitimate experience and I wanna share my feedback for future consideration.