What problem does it solve? Developers working with the Vercel AI SDK often rely on outdated training knowledge, leading to deprecated API usage, type errors, and broken implementations. This Skill forces verification against current documentation and source code, ensuring generated code uses the correct modern APIs like generateText, streamText, ToolLoopAgent, and useChat. ## Core Features & Use Cases - Documentation-Grounded Answers: Searches bundled docs in node_modules/ai/docs/ and ai-sdk.dev instead of relying on memory, preventing deprecated patterns like maxTokens or generateObject. - Agent Building Guidance: Implements the ToolLoopAgent pattern with type-safe consumption via InferAgentUIMessage and useChat, including framework-specific quickstarts. - Error Resolution: Maps common typecheck failures to renamed APIs (parameters to inputSchema, maxSteps to stopWhen) via a curated common-errors reference. - Use Case: A developer asks to add a streaming chatbot to a Next.js app. The Skill checks package.json, fetches current model IDs from the Vercel AI Gateway, and generates code using streamText with toUIMessageStreamResponse and the modern useChat transport pattern. ## Quick Start Ask the AI to build a chat agent with tool calling using the AI SDK and verify the code against the current documentation.