OpenAI has added persistent cloud development environments to Codex that can be accessed from different devices, along with a voice-enabled CLI interface and tools for reviewing code, scanning GitHub repositories, and automatically preparing security fixes.
OpenAI announced a set of expansions for its Codex software agent, most notably transforming cloud development environments from temporary spaces linked to each task into reusable and configurable environments. Developers can access these environments from a computer, phone, or the cloud, with approved settings and permissions and a shared workspace for teams.
The company revealed the updates during the DevDay event and said the new environments were designed to reduce the time needed to start tasks and maintain consistent settings across users. Codex was already able to perform tasks remotely in the cloud, but each task was effectively treated as an isolated environment; the new change adds greater persistence and configurability.
Improving Task Execution and Monitoring
Codex CLI is receiving an update that allows tasks to be started and directed using voice commands. OpenAI is also adding an /agents screen to simplify delegating work and monitoring multiple tasks at the same time, along with improvements for editing instructions, resuming sessions, and using worktrees. The company has also introduced a simpler, clearer terminal interface for handling long-running sessions.
The updates include a new code review experience within the ChatGPT desktop app. Users can read and explore change summaries and ask Codex about potential issues before sending comments to merge requests on GitHub or GitLab. Automatic review can also perform an initial scan while the developer is away from their device.
Scanning Repositories and Preparing Fixes
OpenAI also announced Codex Security Cloud, a set of tools for scanning infrastructure and software repositories. Scans can be run on demand, on a schedule, or when new commits arrive in complete GitHub repositories. According to the company, Codex investigates the findings, removes duplicates, and prepares fixes in the cloud, even when the user's computer is turned off.
The tools include models made available through OpenAI's Daybreak Blue cybersecurity initiative, without submitting a separate request for access.
What Changes in Practice?
This update moves Codex another step beyond being a tool that operates alongside the developer toward a distributed workspace that teams can use and monitor across different devices and processes. The practical benefit lies in task continuity, standardized permissions, and enabling review and security scanning while the developer is away. At the same time, the source does not clarify pricing details, availability limits, mechanisms for approving fixes before merging them, or how to manage the risks of granting an automated agent permissions over real repositories and infrastructure; these are points that require scrutiny before adopting the tools in production environments.
API Updates
The announcements included the Decisions API for making real-time decisions based on user-defined questions and predefined answers, with the interface using the Luna model. OpenAI also updated the Agents API to support computer use and enable the use of Amazon Bedrock Managed Agents to build OpenAI agents that run entirely on AWS.