Add Git-supported QC process guidance

[PallabiniDash]

No description provided.

[kieranjmartin]

Thanks for your first draft. I am a little unsure with this chapter as to how we should approach it; the problem being that QC is of course quite varied and how to approach it can differ from company to company.

I think we were considering maybe leaving this chapter for now as there is a QC working group as well, and it may be better to colloborate with them in future, but perhaps we could include what you have here where we frame it as a suggestion of how QC can be implemented using Git?

[kieranjmartin]

Suggestion: integrate into recommendations section by updating https://github.com/phuse-org/git-in-statistical-programming/blob/main/white-paper/_quarto.yml#L17

and then add some wording around this being a suggested way of adopting QC into git

[kieranjmartin]

could include making use of Copilot code review features: creating skills specific to your standards

[aassuied-ps]

Thanks a lot for this @PallabiniDash! I have one comment:

In Develop and Commit Changes, you're talking about restricted data. I think it should be described again in Retain QC Evidence. In this par you talk about Review comments and responses for instance. These comments and responses should never be retained in the repository if they do contain data information (USUBJID, etc).

I'm not skilled enough to propose a full paragraph about AI agents integrated in a git process, but, what could be said I guess is the following:

• Create mardown skills document that could be used by an AI agent during the Pull Request step could make a pre-QC for specific purposes, such as discrepencies between variable names in datasets and the sponsor reference define/metadata, checks on sponsor's good programming practicies, etc. This kind of workflow can be trigger automatically in case of push or pr.