ktktktkt1234/funf-sosexy-mcp — explained in plain English
Analysis updated 2026-05-18
Let an AI chat client send remote commands to a specific Bluetooth device through MCP.
Relay Bluetooth commands from a server to a device using an Android phone as a bridge.
Deploy a self-hosted MCP server on a VPS with token-based access protection.
Check the last command status sent to a connected Bluetooth device.
| ktktktkt1234/funf-sosexy-mcp | pajkegit/epic-sword-forge | ronit049/find-the-perfect-blinkit-location | |
|---|---|---|---|
| Stars | 7 | 7 | 7 |
| Language | HTML | HTML | HTML |
| Setup difficulty | hard | easy | easy |
| Complexity | 4/5 | 1/5 | 1/5 |
| Audience | developer | designer | pm founder |
Figures from each repo's GitHub metadata at analysis time.
Requires a VPS, a domain, an Android phone for the Bluetooth relay, and a security token before any public deployment.
funf-sosexy-mcp is an unofficial bridge that lets an AI assistant remotely control a Bluetooth-connected personal device made by the FUNF brand, identified over Bluetooth as SOSEXY. It works through MCP, a protocol that lets AI clients call external tools, so an AI chat client can send commands to the physical device through this project's server. The setup has three parts working together. A VPS, meaning a rented cloud server, runs the MCP service that the AI client talks to over HTTPS. An Android phone running Chrome opens a control web page, connects to the device over Bluetooth using the browser's Web Bluetooth feature, and repeatedly checks the server for new commands to forward over Bluetooth. An iPhone or computer can be used to talk to the MCP server itself, but only an Android phone can act as the Bluetooth relay to the physical device, because Web Bluetooth is not supported the same way on those platforms. Deployment involves installing Python dependencies on the VPS, copying server and web files into place, generating a random access token to protect the server, setting up the service to run continuously with systemd, and configuring a Caddy web server for HTTPS. Once running, a person opens the control page on their Android phone, enters the access token, and connects to the device, while separately configuring their AI client to reach the MCP endpoint. The available remote commands let the AI set intensity and mode for a few different functions, stop everything, or check the status of the last command sent, though the README notes this status check does not reflect manual changes made directly on the phone. The project describes its Bluetooth protocol as reverse engineered by the community rather than officially documented, states it only works with this specific device model, and warns that use is at the user's own risk. It requires a security token to be set before deploying it publicly, and is released under the MIT license.
An unofficial bridge that lets an AI assistant remotely control a specific Bluetooth personal device via MCP.
Mainly HTML. The stack also includes Python, HTML, FastMCP.
You can use, modify, and share this freely, including commercially, as long as you keep the copyright notice.
Setup difficulty is rated hard, with roughly 1h+ to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.