The full playbook · v0/plugins/11ai-operator-vercel-ai-gateway/skills/11ai-operator-vercel-ai-gateway-budgets-usage/SKILL.md

11ai Vercel AI Gateway budgets and usage

A budget change can either prevent runaway spend or immediately stop production traffic Resolve the exact project, environment, remote target, permission and cost boundary, and acceptance check before acting.

Version baseline: Use current generally available AI Gateway behavior as of July 2026, including AI SDK 7, multimodal and realtime models, routing, budgets, BYOK, regional inference, and data-retention controls.

Inspect first

npx vercel ai-gateway api-keys --help 2>/dev/null || true
rg -n 'budget|quota|user|customer|feature|AI_GATEWAY' TARGET

Resolve billing team, key, environment, current spend, reset period in UTC, traffic owner, and alerting before changing limits.

Confirm before changing:

Exact key and environment.
Budget amount and reset period.
Per-user or workload attribution.
Expected rejection behavior at cap.

Operate

npx vercel ai-gateway api-keys create --help 2>/dev/null || true
npm test --if-present -- TARGET

Tag requests with low-cardinality identifiers, set caps from measured use and headroom, and handle budget rejection explicitly.

Never create keys, raise or remove budgets, change reset periods, top up credits, or export user-level usage without approval Require explicit approval and preview targets, permissions, counts, cost, remote effects, and rollback.

Verify and report

npm run typecheck --if-present
npm test --if-present
npm run build --if-present

Confirm displayed budget, reset time, test rejection in non-production, attribution, alerts, and rollback to prior limit. Report scope, target IDs, files, remote actions, permissions, cost, checks, observability, and rollback. Hand configuration failures to 11ai-operator-vercel-ai-gateway-troubleshooting and seams to 11ai-operator-vercel-ai-gateway-integrations.