@Jacobsklug@openclaw@VadimStrizheus I think the real struggle is getting this running without Claude OAuth and not spending 5000$/mo on API costs. And if you can't use Claude... then it really gets challenging... that said i can totally see this working with Claude.
I don't get the OpenClaw skills security panic.
I just ask my Agent to build the skill I need. Takes 30 seconds. Tailored to my setup, and zero chance of malware.
The safest skill registry is no registry.
An AI agent that can message across Telegram, Slack, Discord, Signal and WhatsApp from a single conversation. No switching apps, no webhooks, no glue code.
@OpenClaw's cross-context messaging is criminally underrated.
The best model upgrades don't announce themselves.
Opus 4.6 didn't feel different until I noticed I was correcting it less. Explaining less.
That's the real benchmark.
Serial work feels productive.
Parallel work is productive.
The aha moment isn't learning this concept with AI Agents.
It's realizing how much you've been leaving on the table.
@karrisaarinen Ha! We’ve accidentally built this exact workflow over the past few months using Linear. 👍🏻
Our “aha” moment: When we connected Claude (Code) via MCP. Now, every workflow pulls context from Linear: issues, strategy, vision, goals.. you name it. It’s the single source of truth.