What problem does it solve? Integrating multiple AI model providers means juggling separate API keys, model IDs, billing accounts, and failover logic. This Skill guides you through routing all model traffic through Vercel AI Gateway with correct authentication, live model discovery, spend controls, and request debugging. ## Core Features & Use Cases - Setup & Authentication: Configure AI Gateway API keys or Vercel OIDC tokens, migrate existing OpenAI or Anthropic SDK integrations by changing the base URL, and handle BYOK provider credentials correctly. - Model Discovery & Routing: Fetch the live /v1/models catalog, choose models by modality, capability, price, or data retention, and configure provider ordering, model fallbacks, prompt caching, and reasoning effort. - Spend & Observability: Set budgets at team, project, key, or user scope, query credit balances and per-generation costs, and debug requests through AI Gateway Logs. - Use Case: You want Claude Code or Codex to route its own inference through AI Gateway with a spend cap. Run vercel ai-gateway setup with a budget flag, then verify requests appear in Logs. ## Quick Start Ask the agent to add Vercel AI Gateway to my app, fetch the live model list, and make one authenticated test request with a chosen provider/model ID.
