I think I resolved this, it's a bug in ACPX, whenever acp is used, the acp runtime can overwrite the openai-codex oauth, might not be related, but sounds similar to what I was experiencing, randomly getting logged out and model config resetting / token refresh https://t.co/sG6jP8hlST
@steipete PR open with fix that works on my openclaw setup
Introducing Claude Opus 4.7, our most capable Opus model yet.
It handles long-running tasks with more rigor, follows instructions more precisely, and verifies its own outputs before reporting back.
You can hand off your hardest work with less supervision.
@badlogicgames I'm sure you are inundated with posts, tags and mentions.
First wanted to thank you for building such an elegant / eloquent api with Pi, you can see it has taken a lot of thought in terms of naming and functional boundaries etc. I've really enjoyed following you on socials and reading your genuine insights into AI and Agentic engineering.
Here's my question:
I wanted to know how you are running pi on your local machine, and on remote servers in terms of the execution layer, I read your blog and you mentioned that you sandbox everything and give full permissions? If you are doing this, what is the current solution you use to sandbox? Docker Containers? Just curious as I'm learning more to remove permissions for some autonomous workflows, but want to maintain some level of control of execution without having the agent hang up on permissions requests the whole time.
I'm Boris and I created Claude Code. Lots of people have asked how I use Claude Code, so I wanted to show off my setup a bit.
My setup might be surprisingly vanilla! Claude Code works great out of the box, so I personally don't customize it much. There is no one correct way to use Claude Code: we intentionally build it in a way that you can use it, customize it, and hack it however you like. Each person on the Claude Code team uses it very differently.
So, here goes.
I don't find myself using subagents as much as the standard claude code general agent, am I doing this wrong? Should I invest more time into developing custom subagents?
Actual use case for AI in my day:
- Coding Agent (Claude Code) this is 80% of my use case
- Research (Claude & ChatGPT Desktop apps')
- Internet browser / repeatable tasks (Claude Chrome Extension, Perplexity Comet Browser Assistant)
- Reminders / Schedule (ChatGPT tasks)
I'm helping test @friendtech, the marketplace for your friends on Base 🐰
Get the app and search for my Twitter username to be an early holder of my keys 🤗