Skip to content

rich-codex ⚡️📖⚡️

A GitHub Action / command-line tool which generates screengrab images of a terminal window, containing command outputs or code snippets.

PyPI Version


Having code examples in your documentation is a fantastic way to help users understand what to expect from your tool.

Using terminal screenshots is a good way to do this because:

  • 🌈 Coloured terminal output is supported
  • ↔️ You can fit in long lines without scrolling or cropping (images are auto-resized)
  • 😎 They look cool

However, manually generating these screenshots is a pain 👎🏻 Remembering to update them every time you make a minor change means that they can easily get out of date.

Rich-codex automates this process for you. It searches markdown code for images with shell commands or code snippets. It runs these commands and saves a terminal screen-grab at the embedded path.

Typical use cases:

  • 📷 Example CLI tool outputs that automatically stay in sync with your package
  • ♻️ Syntax-highlighted code snippets that are always up to date with your examples/
  • 🤩 Fast and simple images for your docs with minimal setup


  1. 📖 Write some markdown docs, use an image tag with a backtick command inside:

![`cat cat.txt | lolcat -S 1`](img/cat.svg)
2. 🤖 Add a GitHub Action to automatically run the command, generate the image and commit to the repo:

on: [push]
    runs-on: ubuntu-latest
      - uses: actions/checkout@v3

      - name: Install your custom tools
        run: pip install lolcat

      - name: Generate terminal images with rich-codex
        uses: ewels/rich-codex@v1
          commit_changes: "true"
  1. 🌈 Enjoy reading your documentation My cat rainbow

How it works

Rich-codex is a command-line tool that you can run via a GitHub action or as a command line tool. It works with any markdown (including GitHub READMEs).

It collects either commands or code snippets, together with output filenames and configuration options. Commands are run in a subprocess and the standard output & standard error collected. These are then rendered as an image using Textualize/rich.


Rich-codex creates the images that your markdown docs expect. It doesn't require a HTML build-step and doesn't make any changes to your markdown or its output. As such, it's compatible with any documentation engine, including rendering markdown on

Rich-codex needs inputs (commands / snippets) and output filenames to work. These can be configured in four different ways:

  • 🖼 Markdown images
    • Search markdown files for image tags with command alt text. eg: ![`rich-codex --help`](img/rich-codex-help.svg)
  • 💬 Markdown comments
    • Search markdown files for special HTML comments.
  • ➡️ Command-line / action inputs
    • Specify a command or snippet using the action with inputs.
  • ⚙️ Config files
    • Use one or more YAML config files for multiple images and more complex customisation.

Images can be generated as SVG, PNG or PDF (detected by filename extension).