Cointime

Download App
iOS & Android

OpenAI Codex Major Update: Direct Control of Mac Desktop and Automated Task Continuation

According to Beating monitoring, OpenAI has released a significant update for Codex, which is now used by over 3 million developers weekly. The core change in this update is that Codex is no longer limited to writing code; it has begun to engage in the entire software development process. The most notable feature leap is 'background computer control': Codex can now directly view the screen, click the mouse, and type on the keyboard on a Mac, operating any application, with multiple agents working in parallel, each using their own cursor without interfering with the user's ongoing tasks. For developers, this means no longer needing to manually take screenshots to describe iterations of front-end interfaces, as the agents can handle it themselves. A new in-app browser allows users to annotate comments directly on the page to give instructions to the agents, which is suitable for rapid iterations in front-end and game development. The image generation aspect has integrated gpt-image-1.5, enabling the generation of product concept images, UI design drafts, and game materials within the same workflow. Over 90 new plugins have been added, covering development toolchains such as Atlassian Rovo (JIRA management), CircleCI, CodeRabbit, GitLab Issues, Microsoft Suite, and Neon by Databricks. The expansion of automation capabilities is particularly noteworthy. Codex can now reuse existing conversation threads to retain context, autonomously schedule future tasks, and automatically wake up to continue execution, with time spans of up to several days or even weeks. A preview version of the memory feature has also been launched, which can remember user preferences, corrections, and accumulated information from historical operations, allowing subsequent tasks to be completed more quickly. Codex will also proactively suggest to users where to start their day's work or continue previous projects based on project context, connected plugins, and memory, such as identifying pending comments in Google Docs, pulling relevant context from Slack, Notion, and code repositories, and generating a prioritized action list. In terms of developer workflow, new features include GitHub PR review comment handling, multi-device tab management, remote devbox connection via SSH (in alpha phase), and sidebar file previews (supporting PDFs, spreadsheets, slides, and documents). These updates are being gradually rolled out to Codex desktop users logged into ChatGPT. The memory and proactive suggestion features will later be available for enterprise, educational, and EU/UK users, while computer control currently only supports macOS.

Comments

All Comments