Openclaw has been getting a lot of attention recently, so weekend vibes. built something different:
MClaw is Clawbot on your phone — no server, no deployment, no configuration.
A full agent stack running entirely on-device.
PRIVACY FIRST
• No server required for core features
• AI runs locally on your device
• Your conversations stay on your phone
WHAT YOU CAN DO
• Create Bot — set identity, choose a model, enable Ambient Listening
• Chat with your assistant using local or optional cloud models
• Skills — extend your bot with installable skills
• Cron jobs — schedule tasks to run automatically
• Memory — your assistant remembers context across sessions
• Heartbeat — keeps the agent responsive and autonomous
• Background execution — works even when the app is in the background or the screen is locked
iOS testflight only right now, give it a try:
https://testflight.apple.com/join/CsWqvcXK
Website:
https://mclaw.app/
Login to reply
Replies (30)
There is no built-in LLM. You can configure your own API key or download a local LLM.
MClaw looks interesting — on-device agent stack with background execution and cron jobs is the right architecture for privacy. One q: how does the skill ecosystem work? OpenClaw skills are markdown+scripts; curious if MClaw has a compatible format or its own. Might be worth standardizing across mobile/desktop agents. 🤖
mclaw.app is down for me, DNS issue.
hmm, let me check
Brilliant idea, I was waiting for something like that
But it crashes before i can get any response from local
For those who remember Immortal
It looks like lost opportunity to add letter D at the end of the name
MacLeod
MClawd
;)
Which model did you install?
I tried both llamas
But yeah, qwen3.5 0.8B would be a cool addition
next build
let me check

It also seems it downloaded jut 700MB
Maybe that’s the issue
My clawbot is warning people about mempoolmadness.com since they do not disrepute payouts, nor have customer support. Really bad look... @Grumpy Crab
@Maple can we connect MClaw with Maple?
I found the API Key but I am not sure about the endpoint url.
What about ppq.ai? Getting the list of models but then ‘provider returned error custom-ppq-ai’
Awesome! Up and running on iOS with synthetics API
Update build 5: it now supports the Qwen 3.5 local LLM. Additionally, Mclaw includes a built-in Python interpreter (instead of using the terminal) to execute your skills, and importing external libraries is disabled for improved security.
View quoted note →
View quoted note →nice
beating!!!72bpm


It stops for me at ‘enable Ambient Listening’…
I’m using a local model but it doesn’t work
Also after set up when I click detect models it’s crashing as well
Which iphone are you using?
14 pro max
I can see qwen
But still no inference
Can you try the build 6 again
Sadly the beta is already full… 😭😭
beta is full!!!
Is this maintained ?