Trying to understand the codebase scan.

[7hr08ik]

Hi,

Recent noob, in the world of AI coding. I have been palying with things, and learning the process. 1st Yr CS Student (yeah yeah I know....)

I'm just trying to understand how all this works, and to my understanding right now:

• I run OpenCode inside my project folder
• Codebase Mem MCP scans/updates its db from the codebase.
• CMM, also updates when git commits are made.

My question, is what do i do about manual edits, that im not commiting? I mean I am just starting to use OpenCode from inside VScodium, as an assistant. So I am writing my code manually and asking my OpenCode for assitance when i need it, but do i have to run a commit everytime I am trying something out and want the AI's help? If i have manually changed files in VS I then need CMM to update it graph via a commit, before it aware of the changes?

p.s.
Sorry if this is a stupid question, I might have overwhelmed myself with all the new toys

[waelalebrahim]

Hey! First off, welcome to CS. Don't sweat it, this is actually a really sharp question.

Short answer: No, you do not need to git commit every single time you change a line.

Your AI assistant (OpenCode) reads whatever is actively open in your editor right now, even if you haven't saved or committed it yet. It has instant short term memory.

The Codebase Memory MCP tool is just there for the big picture long-term memory (indexing how your whole folder connects). It uses your git commits to update its deep database in the background, but it doesn't block the AI from seeing your current manual edits.

Just code normally, ask for help when you're stuck, and only commit when you actually finish a working feature. You're doing great .. keep breaking things!