OpenAI introduced reusable cloud development environments for Codex at its DevDay event on Tuesday, extending the software engineering agent’s workspace beyond a developer’s laptop. Accessible from any device, the environments make cloud work more persistent and configurable rather than keeping each task in a separate remote sandbox.
Reusable environments for shared development work
Codex already supported running tasks remotely in the cloud. The change is in how those environments are managed: developers can reuse a more persistent, configurable workspace instead of treating each cloud task as an isolated environment.
OpenAI says the reusable environments are designed to shorten task startup and provide teams with a shared workspace using approved settings and permissions. The company describes access spanning a computer, remote use from a phone and work in the cloud.
Voice controls and a new task view in Codex CLI
The Codex CLI update adds voice input for both starting tasks and directing their progress. A new /agents view is intended to make delegating work and monitoring several tasks at once easier.
Other CLI changes address routine interactions, including prompt editing, returning to earlier sessions and working with worktrees. OpenAI is also rolling out a redesigned terminal interface intended to improve readability during longer sessions.
Code reviews inside the ChatGPT desktop app
A new code-review experience in the ChatGPT desktop app lets users bring Codex into the review process. They can inspect summaries, examine code changes and ask about potential problems before sending feedback on a GitHub pull request or a GitLab merge request.
Automatic code reviews also allow Codex to carry out an initial review while the user is away from the computer.
Codex Security Cloud scans repositories and prepares fixes
OpenAI is rolling out Codex Security Cloud, a toolset intended to help developers secure their infrastructure. It supports scanning entire GitHub repositories through three triggers: a developer can request a scan, set a recurring schedule or have scans run when new commits arrive.
After scanning, Codex will investigate the findings, eliminate duplicate reports and prepare fixes in the cloud. That work is intended to continue while the user is inactive, including when their laptop is closed.
The toolset also includes access to models offered through OpenAI’s Daybreak Blue cybersecurity initiative. Developers do not need to submit a separate application for that access.
