Claude Code usage shares plan limits

Anthropic says Claude Code usage on Pro and Max plans draws from the same plan-level limits as regular Claude usage, rather than a separate coding-agent pool. Why it matters: Subscription-based agent work can silently compete with interactive assistant use. Teams need quota visibility before routing long coding sessions through a shared plan budget.

Try this: Document which coding tasks use subscription limits versus API billing, then set a review point for long sessions, large-context tasks, and repeated retry loops.