What is everything-openai-codex?

Everything OpenAI Codex, or EOC, is an open-source workflow system designed to make AI coding agents more reliable and repeatable. Where a bare agent setup gives you a model that can write code, EOC adds structure around it: scoped instructions for different kinds of tasks, reusable skill packs, session memory, quality gates, and adapters that let the same setup work across multiple AI tools. The project organizes work around a repeatable loop: intake, routing, planning, execution, verification, capturing results, and resuming where things left off. Each stage maps to specific files and scripts in the repository. Routing uses skill directories and cross-tool adapters, verification uses a quality gate skill and manifest validators, session state is preserved through status scripts and hook snapshots. EOC ships with a large catalog: 60 specialist agent configurations, 232 reusable skills, 110 rules, 28 hook matchers, 29 install modules, and 75 legacy command shims. The skills cover backend and frontend development, security, machine learning, documentation, and release workflows. Cross-tool adapters make these skills available in OpenAI Codex, Cursor, OpenCode, Gemini, Zed, GitHub Copilot, and Trae. Installation can go through the OpenAI Codex plugin marketplace with a single command, or manually by cloning the repository and running an install script with a chosen profile (minimal, standard, or full). The README warns against mixing both approaches and provides a reset procedure if that happens. A dry-run flag is available to preview what would change before committing. The current public release is v2.0.0-rc.1, which adds a desktop dashboard built with Tkinter, updated catalog metadata, and a Hermes operator workflow for structured multi-step task orchestration. An in-tree alpha runtime written in Rust called ecc2 handles session management and is labeled alpha pending stable packaging. The full README is longer than what was shown.