teemd is a local desktop application: a coordinator that runs a supervised crew of AI builders and reviewers against your own Claude Code / Codex sessions, with an independent review gate before anything is kept. The engine, the review-and-repair loop, and the desktop app surface are built and internally verified. It is not yet publicly available — today it runs only as internal builds the founder tests directly.
Role definitions, model-provider configuration, and evaluation baselines for the coordinated-crew approach are complete.
A real engine coordinates builder and reviewer sessions across two model-provider families, running on your own subscriptions rather than a hosted backend.
Independent review, a cross-provider verification gate, and a repair cycle before anything is kept are built and proven end to end.
The desktop app itself: project browsing and editing, a real terminal and preview, a live view of the crew at work, and a coordinator that keeps context across a session. In active development.
Broader validation, outside testers, and a public release are the remaining steps before this reaches anyone beyond the founder.