
Codex Universal
https://github.com/openai/codex-universal- Category
- Developer Tools
- Rank
- No. 1004Tools index
- Pricing
- Open Source
- Type
- TOOL
- Builder
- openai
- GitHub
- 1.1k stars
- Date
About
Base Docker image used by OpenAI Codex environments — preinstalled toolchains for many languages.
What it does
Codex Universal recreates a close local approximation of a Codex setup environment. Pull the container, mount a project, select supported language versions through environment variables, and enter a configured Bash shell.
Why it's ranked here
It is a practical debugging reference for teams customizing Codex environments. Its broad runtime coverage, version switching, verification script, multi-architecture builds, and published provenance are substantial strengths. The explicit warning that local behavior is not identical limits its value for exact reproduction.
What's good
One container covers Python, Node.js, Rust, Go, Swift, Ruby, PHP, Java, Erlang, Elixir, Bun, Bazel, and common build tools. Several ecosystems include formatters, linters, package managers, and multiple selectable runtime versions. Automated checks exercise the version setup and confirm installed runtimes.
Tradeoffs
Supported versions form a fixed subset, so arbitrary toolchain choices are unavailable. The project only runs the AMD64 build itself, although it publishes both architectures. ARM64 has different Java availability and omits Swift. Most importantly, the maintainers describe the image as similar to Codex, not identical.
How to use it well
Use it when developing or debugging environment customization before moving work into Codex. Mount the project directory, request only documented runtime versions, and test the same setup repeatedly in a disposable shell. It does not provide exact parity with the hosted environment or replace Codex itself.
Technical notes+
Dockerfile starts from ubuntu:24.04, installs system packages, and combines pyenv, nvm, rustup, swiftly, phpenv, and mise for runtime management. entrypoint.sh invokes /opt/codex/setup_universal.sh before launching login Bash; setup_universal.sh switches versions from CODEX_ENV_* variables. verify.sh iterates runtime-version arrays and then checks each toolchain. .github/workflows/build-image.yml builds linux/amd64 and linux/arm64, creates latest and commit-addressed manifests, and publishes build provenance. There are documentation drift signals: README.md lists Bun 1.2.10 while Dockerfile sets 1.2.14, and Dockerfile includes Node 24 and PHP 8.5 beyond the README tables.
Observed
- License
- MIT for codex-universal; bundled components retain their respective licenses.
- Primary implementation
- Dockerfile configuration and Bash shell scripts.
- Distribution
- Published as a pullable container image on GitHub Container Registry.
- Interface
- Docker command line, environment-variable runtime selection, and an interactive Bash shell.
- Platform support
- Builds target Linux AMD64 and Linux ARM64.
- Supply-chain metadata
- Human-readable and SPDX software bills of materials are included; CI publishes build provenance.
Read from README.md, verify.sh, Dockerfile, entrypoint.sh, CONTRIBUTING.md, setup_universal.sh, LICENSES/LICENSE, LICENSES/codex-universal-image-sbom.md, LICENSES/codex-universal-image-sbom.spdx.json, .github/workflows/build-image.yml.
What it can do
Provide pre-configured development environment
Docker container deployment request → Ready-to-use development environment with installed toolchains
Execute code in multiple programming languages
Source code files in various languages → Code execution results and outputs
Compile and build projects
Source code and build configurations → Compiled binaries and build artifacts
Install and manage language-specific dependencies
Package requirements and dependency lists → Installed packages and libraries
Run development tools and utilities
Tool commands and configurations → Tool execution results and generated files
Support Codex AI code generation workflows
AI-generated code and execution requests → Code execution results and runtime feedback
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.