Separate observation from input
A mouse bridge does not necessarily capture the screen or run a model. Anan iPhone Use explicitly leaves those jobs to its client.
ARCHITECTURE / COMPARISON
Compare the approaches.
Compare the path from observation to action, rather than treating every ESP32 board or iPhone agent as interchangeable.
Check the exact chip and phone OS
Classic ESP32 HFP audio, S3 native USB and C3 Bluetooth input serve different purposes. OS requirements also differ across apps.
Read what has actually been verified
A development build, a demo and an author’s device test are different kinds of evidence. Prototype limits remain visible in each profile.
Architecture at a glance
Requirements describe the source revision listed on each project profile. This is not a benchmark or a compatibility guarantee.
| Projects | Project role | Hardware | Runs on | Screen / observation | Input path |
|---|---|---|---|---|---|
| Actell / 执言ja1chen | Screen-control agent | Optional ESP32-S3 with native USB for NCM | Native iPhone app with Swift / Rust backend | CoreDevice screen capture | CoreDevice touch actions |
| AI Passport iPhone Agentlewangdev | Screen-control agent | FoloToy AI Passport / ESP32-C3; optional original S3 board | Mac or Linux server | AirPlay / UxPlay on computer | Bluetooth HID; optional S3 USB HID |
| Anan iPhone Useananhq | Input firmware | M5Stack StopWatch / ESP32-S3; other boards require adaptation | MCP server on board; agent runs in the client | Provided by client, not this firmware | USB HID or experimental BLE input |
| C3 AgentZhaoYangGG | Screen-control agent | ESP32-C3; USB is for power and flashing | Native iPhone app | ScreenCaptureKit with explicit sharing consent | Bluetooth HID mouse and keyboard |
| ESP32 Bluetooth Voice Agentmichaellee8 | Voice experiment | AI-Thinker ESP32 Audio Kit V2.2 / ESP32-A1S with Classic Bluetooth and PSRAM | ESP32 firmware plus a computer-hosted Go bridge | No screen capture; Bluetooth call / communication audio | HFP audio, not mouse or touch input |
| ESP32 iPhone Agent / Handsyonehumanbeing | Screen-control agent | ESP32-S3 with PSRAM and native USB | iPhone app; computer for building and signing | AirPlay → ESP32-S3 → BLE → app | USB HID mouse and keyboard |
| ESP32 iPhone Agent with Midsceneawait-ovo | Screen-control agent | Separately powered ESP32-S3 N16R8; USB-to-UART for flashing | Native iPhone app with a bundled Midscene web runtime | ScreenCaptureKit on iPhone; consent for each sharing session | BLE HID mouse and ASCII keyboard |
| ESP32 iPhone Agent for Windowsgrandexecuti | Screen-control agent | ESP32-S3 with native USB; N16R8 is documented | Windows 10 / 11 computer | UxPlay / GStreamer on Windows | USB HID via the board |
| iPhone Agent Harnesscccat6 | Screen-control agent | ESP32 C2, C3, C5, C6, S3 or classic; board requirements vary | Safari on iPhone / iPad | Browser screen sharing | Bluetooth mouse and keyboard via WebRTC bridge |
| Oh My Phone UsegetGate | Screen-control agent | ESP32-PICO-KIT-1 with Classic Bluetooth | Mac MCP server plus iOS app | ReplayKit screen-sharing app | BLE HID for control; HFP for calling |
| Phone Agentboboabw | Screen-control agent | ESP32-S3; default firmware uses GPIO48 RGB | Native iPhone app | User-authorized full-screen sharing | Bluetooth HID mouse and ASCII keyboard |