wtf is chatbot-template?

This repository is a minimal chatbot template built with Next.js, the AI SDK, and shadcn/ui. It is designed to run on the Vercel AI Gateway, which handles the connection to the AI models. The project provides a starting point for developers who want to create a chat application with streaming responses, tool calling, and web search capabilities. The template includes several features. It supports streaming chat with markdown rendering, allowing responses to appear progressively as they are generated. It includes a tool calling example, where the AI can use tools to fetch information. One tool looks up GitHub repositories and shows stats like stars and forks. Another uses each provider's built-in search tool to search the web, showing a "Searching the web" status and then citations for the results. The template also includes a human-in-the-loop questionnaire, where the model can ask clarifying questions and the user answers them directly in the chat interface. Deploying to Vercel requires no configuration. Vercel deployments authenticate to the AI Gateway automatically, and usage runs on your team's AI Gateway credits. For local development, you need to install dependencies with pnpm and provide a gateway credential, either by pulling a token from a linked Vercel project or by creating an API key in the Vercel dashboard. The model list is defined in a file called lib/models.ts, where the first entry is the default model. The README includes a security warning. The API route that handles chat is public and unauthenticated, meaning every request spends your AI Gateway credits. The route validates the request body, restricts models to an approved list, caps output tokens, and aborts generation if the client disconnects. However, these limits bound a single request, not overall volume. The README recommends adding rate limiting, setting a spend limit, and adding authentication if the chatbot is not meant to be public. Developers can add their own tools by creating a file in the tools directory and a corresponding component in the parts directory. Message types are inferred from tool definitions, so renaming a tool field produces a build error rather than a silent failure.