Hey everyone, I wanted to share a couple of cool @openclaw tips and a big unlock I discovered today while troubleshooting my setup.
If you are running your own OpenClaw, these workflow tweaks will save you a lot of headaches.
Tip #1: Keep context tight with Slack Threads
I use Slack to communicate with R2, my OpenClaw agent. A simple but massive improvement to this workflow is to always use threads.
Instead of dumping everything into one massive ongoing thread (or hitting `/new`, start a new thread for each specific task or project.
Why do this?
Tight Context: It keeps your claw's context window focused strictly on the task at hand.
Better Tracking: It makes it much easier for you to visually track your conversations and separate different streams of work without them bleeding into each other.
Tip #2: Use Tailscale + Codex to fix claw bugs
This is the big unlock.
I have a cron job set up to automatically update OpenClaw to the latest version every night.
This morning, I noticed R2 was acting up. While it started dropping some fun new emojis into my Slack messages (eyes, typing indicators, and fire emojis), it suddenly stopped actually executing commands.
I kept getting an “Exec denied” error and a prompt in Slack to “Approve once” or “Approve all”—a permissions roadblock I hadn't seen before.
Here is how I fixed it using a combination of @Tailscale and Codex:
1. SSH via Tailscale
Instead of physically going to the dedicated Mac running OpenClaw, I used Tailscale to SSH directly into the machine from my main workstation.
Tailscale gives you seamless, secure, live access to your OpenClaw box's file system.
Once connected, I opened the OpenClaw directory right inside VS Code.
2. Use Codex in VSCode