Claude Usage & Cost API turns spend into a workflow metric

Anthropic documents an Admin API for Claude Platform organizations that reports usage and cost by model, workspace, API key, service tier, context window, geography, and server-side tools. Why it matters: Agent experiments become easier to govern when cost can be measured by project, key, model, and tool usage instead of reconstructed from invoices after the fact.

Try this: If an organization uses Claude Platform, prototype a small daily report that groups spend by workspace and API key before expanding automated agent runs.