The graph is the repository.
Kin keeps source code, the connections between its parts, and their history in one repository. Supported tools can work with that repository through Kin's interfaces and ordinary files.
One record, not a copy beside it.
A semantic kernel for code: shared repository state for people and coding agents.
In Kin, source, recorded code relationships, and history are repository state, not a disposable index beside another repository. Kin treats functions, types, and recorded relationships as versioned repository data.
The graph is the repository model, not a separate search index beside it.
Keep the answer in the repository.
Kin keeps code, the connections it finds and change history together. People and agents can inspect that record while supported tools still use familiar files. A connection is a reason to look closer, not a guarantee that a change is safe.
Files are still how you work.
Supported tools can use ordinary files while Kin owns the repository state. Kin's local repository storage and queries run on your machine.
Kin and Git.
Kin has its own version control, with Git import and supported export workflows.
Bring edits made outside Kin into its record before relying on it. Kin and Git maintain separate histories.
People and agents read the same record.
People and supported coding agents can query the same repository record through Kin's interfaces. Kin provides the code foundation beneath your existing agents; an agent runner is optional.