OpenAI adds reusable cloud environments to Codex
OpenAI has introduced reusable cloud development environments for Codex, its software engineering agent, allowing developers to access persistent workspaces from a computer, remotely from a phone or directly in the cloud. The company announced the capability at Dev Day alongside updates to Codex CLI, code review, infrastructure security and APIs.
Codex could already launch cloud tasks for remote work. The new environments make those setups more persistent and configurable instead of treating each task as an isolated remote sandbox. OpenAI said the environments are intended to help tasks start more quickly while giving teams a shared workspace with approved settings and permissions.
Persistent workspaces broaden Codex use
The change shifts Codex beyond a workflow centred on a developer’s laptop. Teams can run the agent where they need it while retaining a configured cloud environment for recurring work. That direction complements Codex enterprise adoption in controlled environments as enterprise adoption depends on making agent-driven engineering usable within controlled team environments.
The refreshed Codex CLI adds voice-based task creation and direction. A new /agents view is designed to help developers delegate work and follow multiple tasks at once. OpenAI also cited improvements to prompt editing, session resumption and worktrees, as well as a cleaner terminal interface for readability during longer sessions.
Review and security automation
OpenAI is adding a code-review experience inside the ChatGPT desktop app. Users can review summaries, inspect code changes and ask Codex about potential issues before providing feedback on GitHub pull requests or GitLab merge requests. Automatic reviews allow Codex to conduct an initial review while the user is away from the computer.
For infrastructure security, the new Codex Security Cloud can scan full GitHub repositories on demand, on a schedule or when new commits arrive. Codex investigates findings, removes duplicates and prepares fixes in the cloud even when a user is inactive and their laptop is closed. The toolset includes models from OpenAI’s Daybreak Blue cybersecurity initiative without a separate application.
API additions extend deployment options
OpenAI also announced the Decisions API for real-time decision-making. It uses Luna to answer developer-defined questions with predefined answers. An updated Agents API now supports computer use and can use Amazon Bedrock Managed Agents to build OpenAI agents that run entirely on AWS.
For businesses, the practical implication is to assess persistent workspace permissions, repository scanning and review hand-offs together, because Codex can now continue engineering and security tasks beyond an individual developer session.