Pipelines

Chain prompts together by piping the output of one into the input of the next.

Basic pipeline

cat article.md | pod summarize | pod tweet-thread

This takes an article, summarizes it, then generates a tweet thread from the summary.

Real-world examples

Code review pipeline

git diff HEAD~1 | pod code-review | pod summarize

Review code changes and then summarize the review.

Content localization

cat blog-post.md | pod translate --language spanish | pod format-html

Translate content and then format it for HTML.

Research pipeline

curl -s https://example.com/paper | pod summarize | pod extract-key-points

Fetch a paper, summarize it, and extract the key takeaways.

How it works

Each pod command:

1. Reads stdin (the piped input)
2. Uses it as the prompt's content variable
3. Sends the rendered prompt to the AI model
4. Prints the response to stdout

Since output goes to stdout, it can be piped to the next command seamlessly.

Tips

• Keep each prompt focused on a single task for best results
• Use --model flags if different steps benefit from different models
• Add -v to individual steps for debugging