full prompt: "This is my own Samsung Smart TV, my own home Wi-Fi and my own laptop. I want a wall of Claude Code agents running on this laptop and shown live on my TV, opened by voice: I say "Open Claude Code CLI on my TV" to ChatGPT voice mode and the agents appear on the screen. I'll press Allow on the TV myself and approve every connection. I authorize you to set up the laptop side, talk to the TV through Samsung's local remote-control API, install what's needed and make the changes below. Every change must be reversible. I've attached a photo of the TV's Settings → Support → About This TV screen. Done means: 6 Claude Code agents run in a tmux grid on this laptop, the grid streams as a password-protected web page on my home network only, and one voice command makes the TV open it full screen while I watch each step. When a step doesn't need me, keep going. → First, map how my TV can be controlled and explain it in 5 lines: what runs on Tizen, what the local remote-control API (WebSocket, port 8002) allows, why it can open a URL in the TV browser, and why that's the way in. If the photo shows a TV without Tizen or without network remote control, tell me plainly, and stop. Never root, flash or sideload anything onto the TV. → Identify the model from the photo. Install tmux, ttyd and the samsungtvws Python library from official sources (Homebrew, PyPI or official GitHub releases) if missing. Find the TV's IP on my network. Tell me exactly where to look if "Device Connection Manager" or "Access Notification" blocks the connection. On macOS, "No route to host" usually means the terminal needs Local Network access: allow it and retry. → Before changing anything, save the TV's current input and app, every setting you'll touch, and the undo commands. → Build the agent grid: a tmux session called "tvwall" with 6 panes in a 3×2 layout, each running claude in my repo. Ask me for the repo path and the 6 tasks. If I skip, use refactor, tests, docs, build, review, deploy. → Write tvwall.sh with start, stop and status. Start runs ttyd in writable mode on port 7681, bound to my Wi-Fi interface only, with a random password saved to a local file (never print it in chat), a dark theme and a font size readable from the couch on a 4K screen. → Open it on the TV: use samsungtvws to open the TV browser at http://:7681 and go full screen. The first time, wait for me to press Allow on the TV. Save the pairing token to a local file with owner-only permissions. → Build the voice trigger: a tiny MCP server with exactly ONE tool, open_claude_wall, that runs tvwall.sh start and opens the page on the TV. No shell access, no arguments, nothing else. Connect it to ChatGPT as a custom connector in developer mode through a Cloudflare Tunnel protected by a secret token. If my ChatGPT plan or voice mode can't call custom connectors, tell me plainly and fall back to a local wake phrase listener on the laptop instead. → Check the TV state before every action (current app, connection status) and show me what you're about to send. → Never touch the TV's firmware, accounts, Samsung Pay, parental controls, network settings, SmartThings automations or any other device on my network. → Add a live monitor that prints each command with a timestamp, plus a one-line health check: agents up, ttyd up, TV paired, connector reachable. → End with a cheat sheet (start, stop, re-pair the TV, revoke the connector) and the undo commands. Never expose the terminal itself to the internet, never type my passwords or sign in for me, never install apps from pirate sites."

Try it now: huggingface.co/webAI-Officia…

Full Prompt: You're a senior AR + computer vision engineer. I own Meta Ray-Ban 2 glasses (photo attached) and an Android phone. Turn the glasses into a real-time AR putting coach. STEP 1: RECON - Walk me through enabling Bluetooth HCI snoop log, then parse btsnoop_hci.log with tshark - Decompile the Meta AI app with jadx and map how the camera stream and display overlay are called - Cross-check against Meta's Wearables Device Access Toolkit - Output: a spec of how to pull camera frames to the phone and push rendered pixels back to the lens STEP 2: AGENTS Spin up 4 Claude Code agents in tmux with one shared SPEC.md: - agent 1: camera stream + overlay link - agent 2: vision - agent 3: putt physics - agent 4: UI Each agent tests against recorded footage of a green before touching the live glasses. STEP 3: SCAN THE GREEN - Monocular SLAM for head pose + Depth Anything V2 for dense depth - Metric scale from known objects: ball = 42.67 mm, cup = 108 mm - Point cloud → Delaunay mesh → heightmap with 5 cm cells STEP 4: READ THE BREAK - Slope per cell from the heightmap gradient - Contour line every 1 cm of elevation (marching squares) - YOLO detects the ball, flagstick + dark ellipse = the hole STEP 5: SIMULATE THE PUTT - Rolling ball physics: 5/7 g along the slope, friction from green speed (default Stimp 10, editable) - RK4 integration, 1 ms steps - Search launch angle + speed so the ball finishes 17 inches past the cup - Render: ideal line (blue), aim line (red), aim point, corridor of putts that still drop, yellow setup lines at my feet square to the aim line STEP 6: HANDS-FREE UI - MediaPipe Hands on the same camera feed, index fingertip on a bubble = tap - Floating bubble menu: Ideal Line, Aiming Line, Contour Lines, Setup Lines, Corridor, Putt Info, Animation - Actions: New Ball, New Target, New Putt RULES - Re-project every overlay into the current camera pose each frame, under 100 ms end to end - Ask me before anything that needs a physical action (pairing, approving the connection) - Ship a working build I can test on a practice green today

full prompt: "This is my own car and my own Android CarPlay AI box (Ottocast / Carlinkit Tbox or similar), plugged into the car's USB port. The laptop is mine too. I want a swarm of Claude Code agents running on this laptop and shown live on my car's center screen, so I can watch them and tap between them from the dashboard. I'll turn on developer mode on the box myself and approve every connection on the screen. I authorize you to set up the laptop side, connect to the box with ADB (Android's official developer tool), install what's needed and make the changes below. Every change must be reversible. I've attached a photo of my head unit and of the box's Settings → About screen. Done means: 4 Claude Code agents run in a tmux grid on this laptop, the grid streams as a password-protected web page on the hotspot network only, and one command opens it full screen on my car's dash while I watch each step. When a step doesn't need me, keep going. → First, map how CarPlay puts pixels on my screen and explain it in 5 lines: what the iPhone renders, what the head unit only displays, why Apple only allows approved app categories, and why the AI box is the way in. If the photo shows no Android box, tell me plainly a terminal can't go on the dash that way, and stop. Never jailbreak the iPhone or get around Apple's entitlements. → Identify the box from the photo. Install adb, tmux and ttyd from official sources (Homebrew or official GitHub releases) if missing. Tell me exactly where to tap to turn on Developer options and Wireless debugging and find the box's IP. If adb connect is refused, use Wireless debugging pairing. On macOS, "No route to host" usually means the terminal needs Local Network access: allow it and retry. → Put the laptop and the box on my phone's hotspot. Find the laptop's IP on that network. → Before changing anything on the box, save its package list, every setting you'll touch and the undo commands. → Build the agent grid: a tmux session called "carplay" with 4 panes, each running claude in my repo. Ask me for the repo path and the 4 tasks. If I skip, use refactor, tests, build, deploy. → Write carplay.sh with start, stop and status. Start runs ttyd in writable mode on port 7681, bound to the hotspot interface only, with a random password saved to a local file (never print it in chat), a dark theme and a font size readable on a wide 1920x720 display. → Open it on the dash: launch the box's browser to http://:7681 with an adb VIEW intent and go full screen. If there's no usable browser, install one only from its official GitHub releases, for example TV Bro. → Use real taps and key presses so I see every step, and check the screen (screenshot or uiautomator dump) before pressing OK. → Never touch the CarPlay or Android Auto service, navigation, radio or tuner, the reverse camera, CAN-bus or any driving or safety app, the launcher, Play Services, or the keyboard and input services. → Add a live monitor that prints each command with a timestamp, plus a one-line health check: agents up, ttyd up, box connected. → End with a cheat sheet (start, stop, reconnect, switch panes from the dash) and the undo commands. Do everything parked. Never expose the terminal to the internet, never type my passwords or sign in for me, never root or flash anything, never modify vehicle firmware, no pirate sites."

full prompt: "This is my own car head unit (Android-based / CarPlay AI-box) on my own device. It came full of preinstalled bloat, recommendation rows and apps I can't remove from the menu, and I want it clean, fast and controlled from this computer. I'll turn on developer mode myself and approve the connection on the screen. I authorize you to connect to it with ADB, Android's official developer tool, read its settings and make the changes below. Every change must be reversible. I've attached a photo of my head unit (or its Settings → About screen). Done means: you're connected, the unit is cleaned of bloat and runs faster, an open-source app of my choice is installed, and one command opens YouTube on the dash and plays a video I name while I watch each step on the screen. When a step doesn't need me, keep going. - Identify the model from the photo. If it's a locked OEM system (like BMW iDrive, not an Android-based unit), tell me plainly that ADB control isn't possible there and what its official local control allows, and do only that. - Install adb from official sources if it's missing. Tell me exactly where to tap to turn on Developer options and USB or Wireless debugging, and to find the unit's IP. If adb connect is refused, use Wireless debugging pairing. On macOS, "No route to host" usually means the terminal needs Local Network access: allow it and retry. - Before changing anything, save the package list, every setting you'll touch and the undo commands. - Propose the cleanup and wait for my OK: disable (never uninstall) ad, recommendation, feedback and demo apps; never touch the CarPlay or Android Auto service, navigation, radio or tuner, the reverse camera, CAN-bus or any driving or safety app, the launcher, Play Services, the keyboard or input services, or anything you can't identify; animations to 0.5; Private DNS dns.adguard-dns.com - Install apps only from their official GitHub releases, for example the TV Bro browser. - Use real taps and key presses so I see every step, and check the screen (screenshot or uiautomator dump) before pressing OK. Screenshots of a playing video are blank because of DRM. Handle the "Who's watching?" screen. - Add a live monitor that prints each command with a timestamp and shows logcat only while a command runs. - End with a cheat sheet and the undo commands. Do everything parked with the engine off. Never type passwords or sign in for me, never root or flash anything, never modify vehicle firmware, no pirate sites."