What leaves your machine
To index a repository, cix has to read it. Here is what is sent to the cix cloud and why.
Your repository's code
cix reads your repository and processes its contents on our servers to build the index. Parsing happens in the cloud, not on your laptop — that is what lets a connected assistant answer questions about your project without you pasting files into its context.
This is true however cix reaches your code. With the GitHub App, cix reads the
repository from GitHub. With cix index --push, the files are read from your
working tree and sent from your machine. Either way the file contents reach the
cix cloud, where they are parsed — the paths differ in where the code is read
from, not in whether it is sent.
Note — --push is not a "code never leaves my machine" mode. If that is
your requirement, neither path meets it today. What --push avoids is
installing a GitHub App and granting it repository access.
Git itself is only ever read. cix does not modify your remotes, branches, commits, tags, or identity.
From your code, cix derives and stores the index your assistant queries: symbols (functions, classes, routes), schema parsed from migrations, dependency relationships, and file metadata. This index lives in a private, account-scoped space tied to your account.
Account and usage data
To run the service, cix also handles:
- Account details — your sign-in identity and plan.
- Repository metadata — which repositories you have connected and their Git state (such as the current commit).
- Usage — counts of the tool calls your assistant makes, for quotas and billing.
What this is used for
Building and serving your index, and running your account. That's it — see What stays private for the limits on how this data is used and how to remove it.
If your organization has stricter requirements
Everything above describes the standard service, and it is the same for every account on it. Some organizations have requirements it doesn't answer on its own — a security team that needs to review how code is processed and stored, retention and deletion obligations to meet, or constraints on what deployment approach is workable at all.
Those are worked out case by case rather than offered from a menu. An Enterprise engagement starts with discovery and a technical and security review, and whatever is agreed is written into the contract — so what applies to your organization is what your agreement says, not what a docs page implies.
If that's your situation, start at Enterprise or get in touch.
Note — until an agreement says otherwise, the standard service above is what applies. Nothing on this page is changed by having a conversation.