Murmur BLE Shell vs WurldCam

murmur turns your phone into a remote control for a Raspberry Pi — no Wi-Fi, no network setup, no port forwarding. Just Bluetooth Low Energy (BLE), directly between your phone and the Pi. HOW IT WORKS Install the open-source murmur daemon on any Raspberry Pi running Linux, pair your phone with it over Bluetooth, and you get: • Shell mode — a real interactive terminal on the Pi, right from your phone. • Agent mode — describe what you want in plain English ("what's eating my disk space?", "set up a static IP") and an AI agent plans and runs the shell commands for you, showing every command and its output as it goes. BRING YOUR OWN API KEY Agent mode is powered by your own Anthropic Claude, OpenAI, or Google Gemini API key — entered once in Settings, stored only in your phone's secure keychain. murmur has no backend server and no account system. The only network calls it makes are the LLM requests you trigger yourself, using your own key. WORKS OFFLINE ON THE PI SIDE The Raspberry Pi never needs an internet connection. It just needs Bluetooth and the murmur daemon running. Good for headless boxes, home labs, and devices you'd rather not put on a network at all. PRIVACY BY DESIGN Everything murmur stores — the Bluetooth pairing key, your API key, an optional per-device sudo password, and saved agent conversations — stays on your phone in the OS-level secure store. One tap in Settings wipes it all. OPEN SOURCE Both the daemon and the app are open source: github.com/nekonihq/murmur

WurldCam turns an iPhone Pro into a posed RGBD capture rig. Every recording contains the colour video, the LiDAR depth map, the camera's position and orientation for each frame, and per-pixel depth confidence — written together into a single file as you record, not stitched together afterwards. Depth is stored losslessly, so the millimetres your sensor measured are the millimetres you get back. Built for people who need the geometry, not just the picture: 3D reconstruction, robotics datasets, SLAM evaluation, NeRF and Gaussian-splat capture, photogrammetry research. • Records to the open wurld format (Matroska/WebM container) or Record3D .r3d • Lossless 16-bit depth, per-frame poses in metres with quaternion rotations • Files stay on your device — no account, no cloud, no analytics • Retrieve captures via the share sheet or the Files app • Read them with the open-source Python and JavaScript tools Requires a LiDAR-equipped iPhone. The support page covers device requirements, retrieving captures via Files, reading the format, why frames get skipped, and where to file issues — so it answers the questions review may probe rather than being a placeholder.