Mastering remote engineering work from your phone | OpenAI Developers

This post details significant enhancements to the agentic execution environment, specifically focusing on the ChatGPT mobile app's role as an engineering control plane. It introduces and elaborates on features like remote host connections, worktrees, goals, side chats, inline code review, queued and steering prompts, attachments, skills and plugins integration within the composer, archived chats, and granular security controls. The post emphasizes a new mental model where the phone acts as a control plane for agentic work, enabling users to start, direct, review, and organize development tasks remotely. It provides practical workflows and patterns for using these features effectively, including managing execution context, performing code reviews, and organizing agent conversations. The introduction of 'Steer' prompts for mid-execution guidance and the distinction between 'Plan' and 'Goal' modes are key technical contributions to agent control.
More from the Agentic Execution Environment trail
- Bringing my LED display to life with GPT-Live-1 and Codex | OpenAI Developers
- Rethinking skills and prompts for GPT-6 Astra | OpenAI Developers
- Meet Rosalind Workbench: Empowering every scientist to be their own research team | OpenAI Developers
- Automating repetitive work at OpenAI with Codex
- Codex as a platform: build on the open agent harness | OpenAI Developers