PRACTICAL ROUTE Run a ready private chatbot with a coding assistant and test whether a follow-up uses the conversation and survives a failed request. Your role: Provide private setup values and test the conversation as a visitor, including a second turn and a failure. 1. Use the supplied project and ask your coding assistant to carry out the implementation. You do not need to write HTML, JavaScript or SQL yourself. 2. Follow the current lesson steps, configure only your own permitted accounts and check the result in the actual page/provider. 3. Keep expected results and observed results separate. Record the exact blocked step if you cannot run a live check. The detailed setup below is available to you and your assistant. Code explanations and extra implementation exercises are optional; important access/privacy checks are still part of the practical lesson. --- REQUIRED SETUP (CODE READING IS OPTIONAL) --- This is a separate practice project. public/ is the website folder; netlify/functions/ contains server-only functions. Install Node and Netlify CLI for local function testing, or deploy this project to your own Netlify account. Put real credentials in server environment variables, never in public/ or Git. No provider request runs just by opening these files. Account steps and real delivery must be checked in your own account. Set GEMINI_API_KEY, GEMINI_MODEL (an available text model from your own AI Studio account) and an unguessable LAB_TOKEN at least 24 characters long. The practice code is entered by the owner, not bundled. Use synthetic messages only: they are sent to Google. Check current account pricing, quota and data terms first. This owner-restricted lab is not a public chat service; add user authentication, durable usage limits, monitoring and a privacy notice before public use. No hardcoded retired model or promised free unlimited tier. This is an in-memory four-exchange browser conversation. Reloading clears it. No database or streaming is included. Test the second turn, not just the first greeting. For exact local commands and private environment names, read run-locally.txt and server-env.example.txt. To inspect current account-visible model IDs without hardcoding a retired name, run: node --env-file=.env list-text-models.mjs This helper lists model metadata and prints names only; it does not request a generated answer. Pick a currently supported text candidate and check pricing/quota before setting GEMINI_MODEL. Current model guide: https://ai.google.dev/gemini-api/docs/models