Ethiopian-Cursor-Community/Group-1-project

# C_GUIDE **Your AI co-pilot for any web app.** Capture your screen or any browser tab, ask in voice or text, and get a step-by-step guide with a moving phantom cursor that points to the exact button to click — refined per step by Gemini. --- ## Table of contents 1. What is C_GUIDE? 2. Features at a glance 3. Architecture overview 4. How the AI providers are used - Gemini (`@google/genai`) - Cursor SDK (`@cursor/sdk`) 5. Per-step refinement (the moving cursor) 6. REST API reference 7. Quick start 8. Project layout 9. Environment variables 10. Hotkey & UI cheat sheet 11. Troubleshooting 12. Extending C_GUIDE --- ## What is C_GUIDE? C_GUIDE is two cooperating clients sharing one backend: | Surface | What it does | | --- | --- | | **Web app** at `localhost` | Shares your screen via `getDisplayMedia`, sends a JPEG frame + your question to Gemini, then animates a phantom cursor over the live preview, refining the target with a **second Gemini call per step**. | | **Chrome extension** (`extension/`) | Drops onto every other tab, scans real interactive DOM elements, asks Gemini which element each step refers to, and pins a phantom cursor + glowing highlight to that element. A small purple ring follows your mouse to show the extension is alive. | Both surfaces share the same hotkey: Ctrl + Shift + L (⌘+Shift+L on macOS). --- ## Features at a glance - 🎙️ **Voice or text input** — Web Speech API in app & extension. - 🖼️ **Screenshot understanding** — Gemini vision analyzes a JPEG of the user's screen. - 🖱️ **Phantom cursor** — Moves over the live preview (app) or pins to real DOM elements (extension). - 🔁 **Per-step refinement** — Every Previous/Next click re-asks the AI _"where exactly for this step?"_, so the cursor stays correct even after the page changes. - 🧭 **Step navigation** — Numbered list, Previous / Next buttons, ←/→ arrow keys. - ✨ **Mouse companion** — Purple glowing ring follows your real cursor on any tab to confirm the extension is …