https://github.com/supermarsx/openai-codex-userscript
OpenAI Codex Super UI Userscript
https://github.com/supermarsx/openai-codex-userscript
codex enhancement greasemonkey injection javascript openai userscript violentmonkey
Last synced: about 2 months ago
JSON representation
OpenAI Codex Super UI Userscript
- Host: GitHub
- URL: https://github.com/supermarsx/openai-codex-userscript
- Owner: supermarsx
- License: mit
- Created: 2025-06-29T13:57:12.000Z (12 months ago)
- Default Branch: main
- Last Pushed: 2025-09-11T11:09:53.000Z (9 months ago)
- Last Synced: 2026-04-30T16:32:55.242Z (about 2 months ago)
- Topics: codex, enhancement, greasemonkey, injection, javascript, openai, userscript, violentmonkey
- Language: TypeScript
- Homepage:
- Size: 394 KB
- Stars: 1
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: readme.md
- License: license.md
- Agents: AGENTS.md
Awesome Lists containing this project
README
# OpenAI Codex Enhancer
[](https://github.com/supermarsx/openai-codex-userscript/releases/latest/download/openai-codex.user.js)
OpenAI Codex Enhancer improves the Codex interface with prompt suggestions, sidebars and extensive customization.
**[Install the userscript](https://github.com/supermarsx/openai-codex-userscript/releases/latest/download/openai-codex.user.js)**
Install Violentmonkey from https://violentmonkey.github.io/get-it/ to run the userscript.
## Features
- Customisable prompt suggestion dropdown
- Settings modal with theme, font and UI toggles
- Repository, version and stats sidebars
- Persistent prompt history stored in IndexedDB
- Automatic update checks
This script reuses Codex's own theme variables so no extra network request is needed.
The dropdown suggestions can be customised by clicking the gear icon next to the
list. A prompt will let you edit one suggestion per line. The updated entries are
saved under the `gpt-prompt-suggestions` key in your browser's
IndexedDB storage, so your changes persist across sessions.
## Settings modal
A floating gear icon is added to the side of the page. Clicking it opens a modal
where you can manage your prompt suggestions and toggle various UI options:
- Switch between Light, Dark and OLED themes.
- Hide the “What are we coding next?” or “What should we code next?” header.
- Hide the “Docs” navigation link.
- Hide the "Settings" button.
- Toggle the repository sidebar that lists detected repositories.
- Toggle the version sidebar that displays branches for the selected repository.
- Enable auto-archiving when a task is merged or closed.
- Import or export your prompt suggestions as a JSON file.
The chosen settings are stored in IndexedDB so they apply whenever the
script runs. By default both repository and version sidebars are visible while
auto-archiving for merged and closed tasks is disabled. These behaviours can be
toggled from the settings modal.
## Prompt history
Click the book icon in the action bar to open the history modal. It lists your
previous prompts up to the limit defined by `historyLimit` (50 by default). A
search box filters entries while buttons let you preview, restore or delete
individual prompts. Use the **Clear** button to remove all stored history.
Entries are kept in the `gpt-prompt-history` store of IndexedDB.
## Theme styling
Theme variables for Light and Dark mode are injected from Codex's own CSS so the interface adapts to your system preference. An additional OLED style is provided with deeper blacks. A minimal fallback style keeps the dropdown readable.
The script locates the ChatGPT prompt input using a set of fallback selectors:
1. `#prompt-textarea`
2. `[data-testid="prompt-textarea"]`
3. The first `.ProseMirror` editor element
If the page structure changes, update the `findPromptInput` function in
`openai-codex.user.js` accordingly.
## Installation
Install the development dependencies with `npm`:
```bash
npm install
```
## Running tests
Automated tests run in Node using [jsdom](https://github.com/jsdom/jsdom).
Install dependencies first to ensure `jsdom` is available, then run the test script:
```bash
npm install
npm test
```
This executes `node test.js` and should print the results of the example DOM
manipulations.
## Development setup
Always run `npm install` before executing `npm test`. The tests depend on the
`jsdom` package, which is listed under `devDependencies` in `package.json`.