Hi everyone,
I’m having an issue connecting the OpenClaw Companion App (Android) to my self-hosted VPS.
Setup:
Ubuntu VPS (Docker 2026.4.21)
Using Claude via gateway
Telegram bot & webchat working fine
Problem:
Companion app connection fails/times out.
Checked:
Firewall (ports open)
Docker running fine (no major errors)
Heartbeat disabled (issue existed before)
Question:
Do I need to configure specific WebSocket ports or gateway.remote.url in .env for remote connection?
Any help would be appreciated 🙏
#: [Help] Companion App failing to connect to VPS-hosted OpenClaw
1 messages · Page 1 of 1 (latest)
it wont connect
- did you paired the app with openclaw??
- since the Android and VPS are on completely different IPs it wont connect and if you find a way to connect it , it will be dissconected very often
Yes, I tried to pair it. I created the remote URL in the terminal and then tried to scan it using the QR code, but the connection still failed/timed out.
Just a heads-up, I am very poor with tech terms, so I might not know the exact technical names for everything, but that is the exact process I followed.
share logs of the app
To give some context on why I am trying to set this up: my main requirement is true, 100% hands-free operation.
I need a setup where I can issue voice commands to my agent (whether through the Telegram bot or the Companion App) completely hands-free while driving, without having to physically touch or wake the phone screen.
Does the Companion App support native voice activation or background voice-to-text routing? Or is there a recommended workflow within the OpenClaw ecosystem to achieve a completely hands-free experience?
i have created a plugin for this but havnt released it yet ...
Is there any other way to do this🤔☹️
tell me what exactly you want and why you want to connect android with Agent?
if it fits inside my plugin then i will release it within 6h and you can be able to use it and control your android ..
That would be absolutely amazing! Thank you for offering.
The user works in the field and drives a lot. He needs to send quick updates, notes, and commands to the OpenClaw agent while driving. For safety, this must be 100% hands-free (no tapping the screen, no looking at the phone).
The Problem: Currently, if he tries to use Android's native voice assistant (e.g., saying "Hey Google, send a Telegram message to Brutus"), Google Assistant blocks it. It crashes because the agent is a Telegram bot and doesn't have a valid phone number saved in the phonebook.
What I exactly want:
I need a way for the Android phone to listen for a voice command, transcribe it, and send that text directly to the Agent/VPS in the background. Whether that is through a persistent listening mode in the app, a custom wake-word, or a way to intercept Google Assistant commands so they route directly to OpenClaw instead of failing.
If your plugin can act as a bridge to send voice-to-text directly to the agent without requiring the user to physically touch the phone screen, it would completely solve our biggest roadblock! Let me know if this fits your scope.
if you limit your desires uptoo this much only , i can make an seperate app for this eaisly .... its not too hard .... and its a bit from my plugin , let me see whether i integrate or create new , ill let you know after any progress on this !!
That sounds absolutely perfect! Yes, limiting it to just a simple 'voice-to-text directly to the agent' is exactly what we need. Nothing complex, just a reliable hands-free bridge.
I really appreciate you taking the time to look into this or building a separate app. Take your time, and please keep me updated on any progress! I would be more than happy to test it out for you as soon as you have something ready