You’ve got a Mac mini setup @openclaw pulled your favorite model from @ollama and spent weeks trying to figure out all the flakiness of crons and whole setup?
sudo pmset -a sleep 0 displaysleep 10
Your openclaw literally falls asleep on the job! This stops it doing so…
Don’t do “understand & update yourself”…run onboarding after update…
Got 2x @openclaw on same Mac and one is always broken and needs love, other is a beast that never skips a beat. Both running on @ollama local models and GLM-5 cloud.
Only difference is how they’re updated.
If you’re running @openai GPT-5.4 in @openclaw and you see this in a response from agent:
Note: I did not schedule a reminder in this turn, so this will not trigger automatically.
Hit that compact button!
I’ve test run GPT-5.4-mini in my @openclaw for less important tasks…oh…what I mistake to make…
Still needed proper GPT-5.4 to nudge it every 2-3 days to get my morning briefing back up to spec…
@steipete My @openclaw figured it out and fixed itself 😊 it was beautiful to watch…a bit annoying as it kept updating me with very step it took, but still…awesome to witness!
Your @openclaw will eat up tokens as it always loads you agents+workplace bootstrap.
Just tell it to stop burning up tokens and it will clean and dedup memory, agents, skills…everything!
You’ll easily go down 30-40%…
After a 3 week daily driving experiment, I prefer GLM-5 over Kimi-K2.5 for local @openclaw via @ollama !
Less eager to jump to action and conclusions. And overall more pleasant personality.
@MatthewBerman Just a simple Telegram mini-app…OpenClaw made it in 5min…
Run fully locally on Mac-mini with ngrok and Qwen3 TTS…
Second thing I’ve built as I spend lots of time in a car and wanted to chat with it hands-free
Openclaw backup opportunity!
1x Mac Mini, 2x @openclaw running under 2 different users. 😊
I’ve “tweaked” mine and it crashed, so my wife messaged it’s to fix and reboot mine! It did a full audit, checked logs, fixed, polished it and got it back up 😊