The headline change is Codex inside the ChatGPT mobile app, with live session state, screenshots, diffs, terminal output, and approvals synced back to the phone. OpenAI also says Remote SSH is generally available, letting Codex connect directly to managed environments. Hooks and scoped programmatic access tokens broaden automation and governance. That combination makes the release useful for both solo devs and teams running repeatable workflows.
This reduces the friction of long-running agent work. Instead of waiting at a desk, a developer can unblock a task, approve a command, or change direction from a phone while the agent keeps moving. For teams, the enterprise angle matters more: it makes Codex fit better into existing environments, permissions, and release processes. That is a real step toward agentic work that stays operationally safe.
Update the ChatGPT mobile app and the Codex macOS app, then connect the machine where Codex is already running. From there, you can inspect active threads, approve actions, and follow the live outputs. If you run managed environments, use Remote SSH plus Hooks to keep Codex inside your existing toolchain. Programmatic access tokens are the piece to use when you want CI or internal automation to interact with Codex securely.
Read Original Post →