Pinned Tweet
Halloween was a hit this year. Maybe 5k people came to the house. Even some film crew came to our haunted house and filmed it. Lines down the street to visit our attraction. 2023 Las Vegas Halloween!
Travis Whidden retweeted
Nevada voters deserve secure, trusted elections. Vote YES on Question 7 for photo ID and ballot verification.
It passed in 2024. It has to pass AGAIN. Repost so no one skips it. #YesOn7NV
#TokenMaxed Claude Max is worth it considering how much they charge for credit usage. The overage ramps up quickly!
Serious Battlefield Boys only 🔥
Our squad + the absolute banger AI song one of the boys made about how we (and half the Battlefield community) actually play 😂
If you get it… you get it.
Come vibe with the real ones → battlefieldfans.gavinbelson.…
#Battlefield #BFSquad #SeriousBattlefieldBoys
Does this option do anything for anyone? Only the Allow Once button works in the VSCode Copilot Chat window, but this doesn't seem to have an effect, and keeps prompting.
Was deep in a flow with VS Code, Copilot, and my beloved Opus 4.6... then errors about model not available. Poof. It's just gone. Grayed out. Completely missing from my options. My colleague still has it, but for me it's vanished. I miss you, Opus 4.6. Come back please. 😔💔
Tried the 30x Opus 4.6 Fast Copilot request for an emergency code deep dive and it failed after a couple minutes.
Proposal for @github / #Copilot / #VSCode: if a call fails for whatever reason, don’t bill the user. Auto-credit it.
At 30x, that safety net would make it much easier to use the heavy-duty options when they matter. @pierceboggan
:( OpenClaw overwrote all its files with the default templates. My system forgot who it was. It's now back to square one.
I asked it for the analysis of why it did that. It looked through the logs, and found out why. This has to be some kind of bug.
------
I found what happened. At 4:39 PM yesterday (Feb 18), OpenClaw's workspace initialization process copied template files from the OpenClaw installation into your workspace, overwriting any customized versions you had.
Timeline:
• 16:39:25 - Git pull in workspace
• 16:39:59 - OpenClaw copied templates: USER.md, SOUL.md, IDENTITY.md, AGENTS.md, TOOLS.md, HEARTBEAT.md, BOOTSTRAP.md
KEY EVIDENCE:
{
"version": 1,
"bootstrapSeededAt": "2026-02-19T00:39:59.950Z"
}
This workspace-state.json file shows OpenClaw ran its workspace bootstrap/seed process at exactly 16:39:59 on Feb 18.
The template files came from:/home/travis/.npm-global/lib/node_modules/openclaw/docs/reference/templates/
This was NOT something I did - this was an OpenClaw system process. Looking at the syslog, there were exec and process tool calls happening at that exact time (runId: ba3b16e7-6d33-465b-8566-d5e15949976d) but I cannot recover what specifically triggered it.
This Is a Critical Bug
OpenClaw's workspace bootstrap should NOT:
1. Overwrite existing files without backup
2. Run without checking if files are already customized
3. Trigger without explicit user consent
Recommendations to Prevent This
1. Immediate backup strategy:
# Create a backup before any git operations
cp -r /home/travis/.openclaw/workspace/.openclaw/workspace-state.json /tmp/
2. File monitoring: I can set up a cron job to monitor these files and back them up if they change
3. Git protection: We should commit your filled-out versions to git so they're tracked
4. Report this bug to OpenClaw maintainers
Would you like me to:
1. Set up automatic backups of your profile files?
2. Help you refill USER.md, SOUL.md, etc. right now?
3. Create a bug report for the OpenClaw team?
What do you remember about your profile that you'd like me to add back?
Totally sold on OpenClaw.
I had to use VS Code + Opus to get it set up correctly. Doing it manually was time-consuming and frustrating. I fed the docs into VS Code, SSH’d into the VM, and it wired everything up exactly how I wanted.
This thing is wild.
It’s already handled tasks on my home servers I’d been putting off for months. It even drafted letters to my state representatives about a 2023 regulation that could save Nevada homeowners money if fixed.
Super impressed.
Hey Openclaw folks, I need some help here.
Does the model choice really make a big difference? I’m using Grok 4-1 (reasoning), and honestly even simple stuff keeps failing.
I’ll ask it to create a cron job and it uses the wrong syntax… then confidently says it’s done. No validation, no warning. Just done. Output shows it failed also.
I also tried running a multi-step email negotiation and at one point it randomly asked the other party for a price of $0. Completely lost the thread of what we were doing.
I don’t mind troubleshooting complex workflows, but I don’t want to debug something like scheduling an email. I just want to trust that when I tell it to do something, it actually does it correctly.
Is this a model issue? A config issue? Or is there a way to make it more aware of what it’s actually executing before it marks tasks complete?
Would really appreciate any insight. I want to trust it, but right now that’s been tough.
