Codex ChatGPT Control
Automates and manages visible ChatGPT web sessions with built-in permission safeguards and structured manual handoffs.
Codex ChatGPT Control enables agents to interact with visible ChatGPT web browser sessions through the codex-chatgpt-control SDK. It supports multi-turn threads, approved file uploads, diagnostic checks, and markdown response capture across Node.js and Python environments. Designed with strict safety postures, it enforces a manual-first policy that creates structured handoff templates when automated execution is not explicitly requested, while halting on captchas, login requirements, and permission boundaries.
Key Features
01Automates visible ChatGPT web browser workflows using Node.js and Python SDKs
02Manages conversational threads, file uploads, and markdown response capture
030 GitHub stars
04Enforces strict manual-first execution policies and structured handoff generation
05Detects and surfaces bridge, login, captcha, and rate-limit blockers cleanly
06Redacts sensitive output run reports by default to protect user privacy
Use Cases
01Sending multi-turn prompts and approved file attachments to active ChatGPT web tabs
02Running multi-model review and validation tasks between Claude Code and ChatGPT
03Generating standardized manual prompt handoff templates when browser bridges are unavailable