Repo Intelligence is a repo-first retrieval layer for developers and AI agents. Public beta is intentionally narrow: bring one repository through GitHub, then prove the API or MCP results against real work in your own codebase.
Connect GitHub, pick one repo, and track the submission until it becomes ready.
Call the hosted API once your repo is active and compare ranked files with what you really open or edit.
Use the same repo through MCP when your client or agent already speaks MCP.
The control plane is the public repo-owner path. Connect GitHub, select one supported repo, and activate it into onboarding.
There is no per-repo admin review in the normal path. Intake can still be paused globally, and runtime work still happens behind the scenes.
When the repo reaches ready state, the control plane shows repo-specific API and MCP next steps so you can prove the claim on your own code.
One public beta story across submit, API, and MCP, with a ready-state path that tells users exactly how to validate the repo once it is served.
No billing, no org/team RBAC, no admin redesign, no broader MCP tool surface, and no engine-behavior change in this slice.