What problem does it solve? Developers working with the Vercel AI SDK often rely on outdated training knowledge, leading to deprecated APIs, wrong parameter names, and broken type checks. This Skill forces verification against current documentation and source code before writing any AI SDK code. ## Core Features & Use Cases - Up-to-date API guidance: Searches bundled docs in node_modules/ai/docs/ and ai-sdk.dev instead of relying on memory, covering generateText, streamText, ToolLoopAgent, embed, and tools. - Migration and error resolution: Maps deprecated patterns (maxTokens, maxSteps, parameters, generateObject, toDataStreamResponse) to their current replacements via a common-errors reference. - Type-safe agent building: Guides creation of ToolLoopAgent agents with InferAgentUIMessage for end-to-end type safety with useChat in React. - Use Case: A developer asks to add a streaming chatbot to a Next.js app; the Skill checks the installed ai package version, fetches current model IDs from the Vercel AI Gateway, and generates code using toUIMessageStreamResponse and DefaultChatTransport. ## Quick Start Ask the assistant to build a streaming chat endpoint with the AI SDK and have it verify the current API against the installed ai package docs.