What you pay for

Model usage is never one of them. You bring your own provider keys, so inference is billed to you by the provider at cost — the plan covers the orchestration around it.

A seat is a chair, not a subscription

A seat is one role in one room: Drafter, Reviewer, or Indexer. You can reconfigure which model occupies a seat as often as you like — swapping a model into the Drafter chair does not consume anything. Your plan caps how many seats can be filled at once, not how many roles you may define or switch between.

The plans

• Basic — free forever. Two seats, one active project, 5 MB of workspace storage. Enough to see the mechanic, though two seats cannot break a tie.
• Basic Plus — $14/month. Three seats: the first tier where majority vote actually works, and where the orchestra starts.
• Pro — $20/month. Four seats and a synced knowledge layer.
• Pro Plus — $24/month. Five seats — the full canonical role set.
• Max — $30/month. Six seats, and room for a custom role per project.

Full per-plan detail lives on the pricing page. Annual billing is two months free on every paid plan.

If you cancel

Nothing local is taken away. Your keys stay in vscode.SecretStorage, your PLAN.md and memory/ files stay in your repo, and the extension keeps working at Basic limits. Seats over the new cap become unseated rather than deleted, and nothing already written is locked.

What we never bill for

• Tokens. Your provider bills you directly. We never proxy a request, so we could not meter it if we wanted to.
• Your plans and trails. PLAN.md, EXECUTION.md and memory/ are files in your repository. The storage allowance on each plan covers the workspace we hold on your behalf, not those.