This is a casual, plug‑and‑play template for an AI chat app. Fork it, run it, change a few things, and you’ve got an instant API interface you can use for a project, a demo, or just for fun. Bring your own key, or wire it to your own server. Enjoy.

• Instant UI: Clean chat interface out of the box
• Bring your own key: Paste your Gemini API key (stored locally)
• Markdown answers: Tables, lists, code blocks, links handled nicely
• Take notes: Save outputs and export as PDF
• Tweak the prompt: Tone, length, level, language
• Flexible: Use client key, server key, or both

Clone your fork and install deps (client + server):

git clone <your-fork-url>
cd tufexai/turfex/Server && npm i
cd ../Client && npm i

Run the server (port 3001):

cd tufexai/turfex/Server
# If you want a server fallback key, create .env with:
# GEMINI_API_KEY=your_server_side_key
npm start

Run the client (Vite dev server):

cd tufexai/turfex/Client
npm run dev

Open the app ( http://localhost:5173).

• Click the settings icon in the chat header
• Paste your Google Gemini API key and press Add
• Your key is stored in the browser (localStorage) and used directly from the client
• You can delete the key any time; the server fallback key (if configured) will be used instead

Grab a Gemini key from https://aistudio.google.com/app/apikey.

• Where: turfex/Server/index.js, helpers in turfex/Server/utils/*
• Do what you want: change routes, add endpoints, call any provider, add auth/rate limits
• Return whatever shape you like; the client just needs text content for the chat
• Edit System Prompts

Default API we ship:

POST /api/turfex
{ messages, tone, length, level, language, apiKey? }

Behavior:

• If a client key is present, use it
• Otherwise use the server key (if set)
• Uses @google/genai with gemini-2.5-flash by default

• Where: turfex/Client/src
• Edit models and actions: components/chat-input.jsx
• Response rendering (Markdown, buttons): components/chat-response.jsx
• Markdown styles/components: components/markdown.jsx
• Main page & modals: pages/Chat.jsx
• API call: api/turfex.js
• Local key store (Zustand): stores/apistore.js

Tip: The Markdown renderer is already mobile‑friendly and supports code blocks, tables, images, and safe links.

turfex/
  Client/                 # React app (Vite + Tailwind)
    src/
      components/
        chat-input.jsx     # Prompt bar with model selector and actions
        chat-response.jsx  # Rendered responses (Markdown + buttons)
        markdown.jsx       # Markdown renderer (GFM, responsive)
      pages/
        Chat.jsx          # Main chat page (notes modal, settings modal)
      api/turfex.js       # Client -> Server API call
      stores/apistore.js  # localStorage-backed API key store (Zustand)
  Server/
    index.js              # Express server, Gemini call, server key fallback
    utils/BuildPrompt.js  # Prompt builder (tone/length/level/language)

Server (turfex/Server/.env):

• GEMINI_API_KEY – optional. If provided, the server will use it when the user hasn’t added a client key.

• All UI is plain React + Tailwind. Update components in src/components and src/pages.
• Swap icons (Lucide), tweak colors, spacing, and typography.
• Replace the model list in ChatInput.jsx with your own choices.
• Edit the settings modal content in pages/Chat.jsx.
• Replace the gradient/glow around the input or remove it entirely.

Tip: The Markdown renderer (components/markdown.jsx) already includes mobile-friendly typography, code scrolling, tables, images, and safe links.

• From each response, you can save text into Notes
• Notes are persisted locally and can be exported to PDF (jspdf)

The client builds a structured prompt (tone, length, academic level, language) and calls the server:

POST /api/turfex
{ messages, tone, length, level, language, apiKey? }

• If a client API key is provided, the server uses it.
• Otherwise, it falls back to the server key (if set).

The server uses @google/genai to call gemini-2.5-flash and returns clean text.

• Client: any static hosting (Vercel, Netlify). Build with npm run build.
• Server: any Node host (Render, Railway, Fly.io, VPS). Ensure GEMINI_API_KEY is set.