Files
llmx/codex-cli/examples/README.md

45 lines
2.1 KiB
Markdown
Raw Normal View History

# Quick start examples
This directory bundles some selfcontained examples using the Codex CLI. If you have never used the Codex CLI before, and want to see it complete a sample task, start with running **camerascii**. You'll see your webcam feed turned into animated ASCII art in a few minutes.
If you want to get started using the Codex CLI directly, skip this and refer to the prompting guide.
## Structure
Each example contains the following:
```
examplename/
├── run.sh # helper script that launches a new Codex session for the task
├── task.yaml # task spec containing a prompt passed to Codex
├── template/ # (optional) starter files copied into each run
└── runs/ # work directories created by run.sh
```
**run.sh**: a convenience wrapper that does three things:
- Creates `runs/run_N`, where *N* is the number of a run.
- Copies the contents of `template/` into that folder (if present).
- Launches the Codex CLI with the description from `task.yaml`.
**template/**: any existing files or markdown instructions you would like Codex to see before it starts working.
**runs/**: the directories produced by `run.sh`.
## Running an example
1. **Run the helper script**:
```
cd camerascii
./run.sh
```
2. **Interact with the Codex CLI**: the CLI will open with the prompt: “*Take a look at the screenshot details and implement a webpage that uses a webcam to style the video feed accordingly…*” Confirm the commands Codex CLI requests to generate `index.html`.
3. **Check its work**: when Codex is done, open ``runs/run_1/index.html`` in a browser. Your webcam feed should now be rendered as a cascade of ASCII glyphs. If the outcome isn't what you expect, try running it again, or adjust the task prompt.
## Other examples
Besides **camerascii**, you can experiment with:
- **buildcodexdemo**: recreate the original 2021 Codex YouTube demo.
- **impossiblepong**: where Codex creates more difficult levels.
- **promptanalyzer**: make a data science app for clustering [prompts](https://github.com/f/awesome-chatgpt-prompts).