GitClear offers various options to control the extent to which source code can be anonymously analyzed or submitted to third parties. This page describes the options available for ensuring your code is handled with the desired level of permissions.

linkSharing code with OpenAI

Each customer can decide whether they want to allow OpenAI to summarize the changes that are happening within their repo. We offer three options for what can be shared with OpenAI:

It is possible to share no data, share only the most basic data, or share code line data

Why do we recommend sharing code lines? OpenAI doesn't use queries submitted via API for training. Here is their official documentation where they state: "Please note that we do not use content submitted by customers to our business offerings such as API to improve model performance."

linkWhat data is sent to OpenAI?

GitClear does not store full code files in its database, nor does it send full files to OpenAI, unless the full file was created in a single commit. Our method of summarizing a diff is to send a snippet of changed code and ask for it to be summarized. This means that no individual query is likely to have more than 5-10 changed code lines, nor will the query know the file name/path of these changes.

If the line is not changed within your git repo, it won't be sent for summarization.

If "Basic OpenAI data sharing" is selected, then we won't send individual code lines, but we will send commit messages, such that we can succinctly describe the high-level story of what changed across a group of commits.

linkAllowing support access

In the main "Settings" tab for an entity, you can control whether it's possible for GitClear staff to simulate logging in to your entity.

If you choose "No," then GitClear team members won't be able to view reports from your GitClear account. This prevents debugging many types of support questions.