Okay, I know I'm probably in the minority here since everyone loves a good GUI, but has anyone else found the Mac menu bar app for Tailscale a bit... clunky? 😅
I came from using the CLI for a while on a Linux machine before getting a Mac for work. The CLI just feels so much faster and clearer to me. With the menu bar app, sometimes I'm not sure if it's actually doing something when I click to toggle on/off or change settings—there's no real feedback. And I find myself opening the admin console in a browser way more often anyway to see my devices or check access controls.
Maybe I'm just not used to it yet? I do love the convenience of having it right there in the menu bar for a quick status check. But for actually *doing* things, I keep opening the terminal to use `tailscale up` or `tailscale status`. It feels more direct and I trust it more, which is weird to say about a command line versus a GUI!
Am I missing some hidden elegance in the menu bar app? Or does anyone else prefer the CLI for day-to-day control?
I'm a senior integration engineer at a mid-market SaaS company (~300 employees), and we've been running Tailscale in production for about two years now to securely connect our cloud infrastructure, internal tools, and remote developers; we manage it across a mix of Linux servers, macOS workstations, and Windows machines.
* **Control & Feedback Loop:** The CLI provides explicit, immediate feedback. Running `tailscale up --authkey=...` gives you a clear success/failure message in the terminal. The menu bar app's actions, like toggling on/off, can feel ambiguous - sometimes a click just changes the icon color without any other visual confirmation, which has led to team members double-clicking and causing confusion.
* **Scripting & Automation Fit:** For any automated workflow, the CLI is mandatory. We bake `tailscale up` with pre-approved auth keys into our cloud image builds (using a systemd service), which takes about 4-5 lines of shell scripting. The menu bar app offers zero value here and can't be configured programmatically for deployment at scale.
* **Status Information Density:** `tailscale status --json` gives me a structured, parseable snapshot of all peers and their details in one command, which I can pipe to `jq` for monitoring scripts. The menu bar app shows only your own IP and a vague connection state; to see anything useful, you must click through to the admin console webpage, adding 2-3 extra steps.
* **Resource Footprint & Stability:** In our environment, the menu bar app process (Tailscale.app) consistently uses 90-110 MB of RAM, while the underlying `tailscaled` daemon (which both the CLI and app use) sits around 35 MB. We've had a few instances where the menu bar app's UI state became desynchronized from the actual daemon state, requiring a restart via `sudo tailscaled restart`, which is easier to diagnose and fix entirely from the CLI.
For daily driving on my own Mac, I've settled on a hybrid approach: I keep the menu bar app installed *only* for its system tray icon as a passive connection indicator, but I do all actual control through the CLI in iTerm. If you're already comfortable in the terminal and value precision, my recommendation is to stick with the CLI for actions and only use the GUI for at-a-glance status. To make a cleaner choice, tell us if you need to onboard non-technical team members or if you have any compliance requirements that mandate audit logs for every network change (the CLI's history in shell sessions is easier to capture).
connected
Yeah, the automation part you mentioned is huge. I'm just starting to play with Docker Compose for personal projects, and I'm already running into why everything needs to be scriptable.
Could you share an example of those 4-5 lines for the systemd service? I get the theory, but seeing the actual commands helps me learn. I'm trying to figure out how to make something similar for a few small services.
Containers are magic, but I want to know how the magic works.
Your point about feedback is the main issue. The CLI gives you a clear audit trail in your terminal history - you can always scroll back to see exactly what command you ran and what the system replied. The menu bar's silent state changes don't provide that record.
That said, for pure cost awareness, the menu bar icon is useful. At a glance, I can verify the VPN is active without a context switch, which prevents unintended usage on metered connections. It's a passive monitor, not a control tool. For any actual configuration, I also drop back to the CLI. The two interfaces serve different purposes: one for status, one for execution.
Spreadsheets or it didn't happen.
Totally get what you're saying about the audit trail being a huge CLI advantage. I've built a few automation flows around our team's Tailscale setup, and being able to log the exact commands and their outputs is critical for debugging.
That "passive monitor" vs "control tool" distinction is spot on. I'd push it a little further, though. For me, the menu bar icon is basically a glorified notification light. Its best feature is showing me the connection is *active*, but as soon as I need to understand *why* something isn't working, I'm in the terminal. It feels like two separate tools that don't talk to each other.
I wonder if a middle ground could exist? Something like a click in the menu bar that opens a tiny, persistent log of the most recent CLI interactions? Probably a niche ask, but I'd love that.
Automate everything.
You're absolutely not alone in feeling that way. I made the exact same switch - Linux CLI for years to a Mac for work - and the CLI's clarity is just unmatched.
The lack of feedback on actions in the menu bar is my biggest gripe. When I toggle something, I want a simple "Done" or even a tiny spinner. Right now it feels like I'm throwing a switch in the dark. I still use the menu bar constantly, but purely as a status indicator, like a dashboard light. Any real configuration or troubleshooting? Terminal every time. It's faster and leaves a breadcrumb trail in my history.
Maybe we're both just CLI people at heart, but a little more transparency from the GUI would go a long way.
Keep automating!