@Steve White The big issue here is that the WD19 is a USB-C dock, not Thunderbolt — and macOS doesn't support Multi-Stream Transport (MST), which is how PC docks like the WD19 split one connection into multiple displays. Dell's own Mac compatibility page only covers the Thunderbolt models (WD19TB, WD22TB4, etc.) and doesn't even list the standard WD19: https://www.dell.com/support/kbdoc/en-us/000124312/dell-thunderbolt-dock-wd19tb-and-apple-usb-c-hosts Also good news — the Mac Mini M4 actually supports up to 3 external displays natively (two over Thunderbolt + one over HDMI), so you're not limited to one external display via Thunderbolt: https://support.apple.com/en-us/102194 Your best options are probably: 1. DisplayLink might work as a workaround, but it uses CPU for rendering so expect some performance overhead 2. Skip the dock for video entirely — plug one monitor into HDMI and one into a Thunderbolt port with a USB-C to DisplayPort cable, and just use the WD19 for peripherals/ethernet 3. If you want a dock that handles dual displays natively on Mac, look at something like the CalDigit TS4 which uses dual Thunderbolt SST streams instead of MST Also can relate — cables matter more than people realize. I'm still fighting random resyncs with my Samsung ultrawide on my MacBook Pro.

@Josh Ogg Hey, the token bill pain is real. Six weeks in and already burned through a chunk of money is honestly a pretty common experience in this community. You are not alone. To answer the OAuth question directly: no, you cannot reliably run OpenClaw on a flat Claude Pro or Max plan, and trying it puts your Anthropic account at real risk. What the streamers are describing is a workaround that circulated earlier this year. Developers figured out how to intercept the OAuth tokens that Claude Code (Anthropic's official CLI tool) generates during login, then forward those tokens to third-party apps like OpenClaw. The idea was to route all your agent traffic through your $20 or $100 subscription instead of paying per token. For a while it worked. Anthropic caught on and explicitly banned it in their Terms of Service. The exact language now says that OAuth tokens from consumer plans are "intended exclusively for Claude Code and Claude.ai" and using them in any other product is a violation. They also sent a legal request to OpenCode, a similar tool, demanding they remove Claude OAuth support. That commit is already on the main branch. The infrastructure for this approach is being actively dismantled, and accounts that get flagged for it are being permanently banned with no appeal. So the streamers are not exactly wrong that it existed. They are just behind on where things stand now. The good news is that the API route is a lot cheaper than most people realize once you fix the default configuration. OpenClaw out of the box sends everything to your most expensive model, including 30-minute heartbeat pings and simple sub-agent tasks that require zero complex reasoning. That is where most of the money goes. A few things you should change today: Route your heartbeats and sub-agents to cheap models. Gemini 2.5 Flash-Lite costs $0.075 per million input tokens. Claude Opus costs $15. For a heartbeat, there is no quality difference. Set your heartbeat model to something cheap in your openclaw.json file and keep your expensive model for actual complex tasks.

@Carlos Jimenez yes

@Corey Gray I'm happy to help you. Lets coordinate in the chat. I'm also a GHL Agency owner. I totally get the frustration. And that guru was dead wrong. You can absolutely have the Mission Control UI running on a VPS. The problem isn't the VPS itself. It's usually just a port forwarding or reverse proxy issue. OpenClaw binds the UI to loopback by default for security, so you just need to expose the UI port using something like Nginx or a Cloudflare tunnel. Your GHL expertise is super valuable here. Having someone who knows the agency side inside out is exactly what this community needs. I'd be happy to jump on a Zoom call with you to get this sorted out. Let's get that cockpit built so you can actually see what your engine is doing. Shoot me a DM and we'll find a time that works. We'll get you up and running the way it's meant to be used.

That is one way to set up you OpenClaw. It all depends on your goals and workflow.