Hey everyone,
I’m excited to share a project I’ve been working on: Codexy, a Python implementation of the official OpenAI Codex CLI!
Inspired by the capabilities of the original TypeScript Codex CLI, I wanted to bring a similar terminal-native, agentic coding experience to the Python ecosystem. Leveraging AI assistance (fitting for an OpenAI tool!), I was able to put together this initial version in roughly two days.
What is Codexy?
Codexy aims to replicate the core functionality of the Codex CLI using Python libraries like textual
for the TUI, click
for the CLI framework, and the official openai
Python library. It lets you interact with OpenAI models directly from your terminal to:
- Get code explanations
- Generate code edits (via a patch format)
- Execute shell commands (with configurable approvals)
- Read and list files in your project
Why Python?
- To offer a familiar stack for Python developers.
- To explore building TUIs with the excellent
textual
library. - To make it potentially easier to integrate into Python-based development workflows.
Current Status & Disclaimer
Codexy is highly experimental and under active development. It’s a reimplementation and currently lacks some features of the original, most notably the robust platform-specific sandboxing (like Seatbelt on macOS). While it has approval modes, commands in full-auto
currently run without the CLI’s own sandbox. Please use auto-approval modes with caution!
Think of it as an alpha/beta release – expect bugs and rough edges!
Get Involved!
I built this in the open and would love community feedback and contributions!
- GitHub Repo: https://github.com/AndersonBY/codexy
- Installation:
pip install -U codexy
- Quickstart: Check out the README
Please give it a try, file issues for bugs or feature requests, and consider contributing if you’re interested! Let me know what you think.
Cheers,
MakerBi