How openai/codex-universal works
Project type and purpose
• This is a Docker‐based reference implementation (“codex‐universal”) that builds and configures a base image for OpenAI Codex.
• Its primary goal is to provide a local image that mirrors the Codex environment (multiple language runtimes, tools, and entrypoint scripts) for development and debugging.
File structure
Top‐level files and directories:
• .github/workflows/build-image.yml – GitHub Actions pipeline that builds & publishes the image
• Dockerfile – instructions to install runtimes and tools
• entrypoint.sh – container entrypoint that invokes setup
• setup_universal.sh – script that reads CODEX_ENV_* variables and installs language versions
• LICENSES/ – licensing and SBOM outputs
• README.md – usage, supported versions, included packages
• setup scripts and CI config imply a CI/CD → Docker image → registry → user workflow
README for additional insights
• Describes how environment variables (CODEX_ENV_PYTHON_VERSION, etc.) drive runtime installation
• Lists supported versions and extra tools (pyenv, corepack, bazel, etc.)
• Shows usage: docker pull + docker run with env vars, volume mount, working directory
Final recommendations
• Be exhaustive: break the container internals into sub‐components (entrypoint vs. setup vs. individual installers).
• Step with key file names (e.g., build-image.yml, Dockerfile, entrypoint.sh).