@elonmusk@xAI@grok@GrokInsider
Grok Build is impressive, but I'd love to see a full chat-based interface in addition to the CLI.
Natural conversation, long-term context, project memory, and agent collaboration would make it much more accessible for non-developers and power users alike.
The future isn't commands. It's conversation.
Since June 4, 2026 (JST), Computer Use, Browser, and Chrome plugins have disappeared for me in Codex Desktop, and they are also missing on mobile. Restart, logout/login, repair/reset, and reinstall did not restore them, so this does not look like a simple local cache issue.
I also found multiple similar public reports on GitHub:
#25391 https://t.co/Dt0jbih5Ri
#25512 https://t.co/vMFgKX6BVO
#25488 https://t.co/K2mk09TuKc
#25253 https://t.co/zgLdmGaQDY
#25220 https://t.co/Ds2fBeCQEs
#21440 https://t.co/frS83Wsm5d
#24736 https://t.co/ayItjNbKCx
Not posting this as criticism, only to help with pattern-matching and triage. If others are seeing the same thing, sharing your version / OS / date noticed may help.
@OpenAIDevs
Sharing this in case it helps other users and the Codex team identify a potential issue.
Environment: • Windows 11 • Codex v26.601.21317 • ChatGPT Plus • Japan
On June 3, Computer Use and Chrome-related functionality were working normally.
After updating Codex Desktop and installing the Codex Chrome extension, the following appeared simultaneously:
• Computer Use plugins unavailable • In-app browser plugin unavailable • Chrome-related plugin missing
I have already contacted OpenAI Support and provided diagnostic information.
I am not posting this as a complaint. My goal is simply to determine whether other users are seeing the same behavior and to contribute information that may help identify the root cause and improve the product.
If anyone has experienced something similar, I would appreciate hearing about your environment and findings.
#Codex #OpenAI #ComputerUse #AIAgents
@OpenAIDevs
Grok Build tip of the day: check your work!
Just like you would check your work after writing code, use /check-work to have a subagent check on Grok’s code. It will look at the repo, trace alternative code paths, and fix bugs before reporting back to Grok what it found.
Mi entrenador en el gym me preguntó a qué me dedico.
"Sistemas de trading. Open-source."
Se rió. "¿Como day trading?"
"No. Cualquiera puede ver cómo funciona. Todo el código es público."
Soltó las pesas. "¿Entonces te pasas el día mirando código?"
No discutí. Abrí mi laptop.
Una wallet que estaba siguiendo rindió un 71% en un solo mes. Otra movió $4.1M en volumen. Y otra convirtió $500 en $8,200.
Su cara cambió.
"¿Eso... sale de un código?"
Exacto.
Entonces le mostré los repos. Todos públicos. Todos gratis.
El primero: https://t.co/OOcivVredj
Cada operación jamás realizada. Más de 86M de trades. Gratis para descargar. El snapshot te ahorra más de 2 días de investigación.
El segundo: https://t.co/raqPAgzqsc
Bot de market making. Ambos lados del libro. Optimizado en gas. Capa de ejecución en Google Sheets.
El tercero: https://t.co/8JRAQigcLA
ML + heurísticas. Detectó esa wallet de $500 → $8,200 antes de que nadie se diera cuenta.
Market making automático con LLM. Open source del equipo de Polymarket.
Mi entrenador se quedó callado. Luego preguntó: "¿Mi hijo puede usar esto? Le gustan las computadoras."
Perfil: https://t.co/qhxUfIioy1
Le dije que sí. Y le mostré cómo empezar.
La mayoría cree que necesitas un título en finanzas o un fondo de inversión.
No lo necesitas. Solo tienes que leer el código y empezar.
Grok Build 0.2.7 is now out, with /usage, /login, shared terminals across subagents, and improved image understanding
See all updates at https://t.co/G6Q2RYPrmx
We spent a ton of time making worktrees actually work well for agent swarms and large repos.
When you’re running 10s of agents that ship, you quickly realize you need worktree, but git’s defaults are brutal at this scale.
Slow creation, every agent copying the whole repo, can’t check out the same branch twice.
So we fixed those problems in Grok Build.
Now agents can spin up, reset, and go wild constantly. Worktrees are faster than normal Git, everything shares the same base, and it doesn’t eat your disk or grind your SSD.
Try it:
grok -w <label> → new session in a worktree
grok -w <label> -r <session_id> → resume one
grok worktree ... → manage them
SpaceX has almost finished writing V1.0 of an in-house AI training stack in C that exact-maps to 220k GB300s with 800G NICs, making heavy use of pipeline parallelism and getting as close to bare metal as possible.
The potential speed improvement vs JAX for large training runs is over an order of magnitude.