Chaining Prompts

Breaking Complex Tasks into Steps

Some tasks are too complex for a single prompt. Prompt chaining breaks them into sequential steps where the output of one prompt becomes the input of the next.

When to Chain

• The task has distinct phases (extract → analyze → format).
• A single prompt would be too long or complex.
• You need to validate or transform intermediate results.

Design Principles

1. Each step should have a narrow, well-defined task.
2. The output format of step N should be designed as input for step N+1.
3. Use XML tags to structure the handoff between steps.
4. Validate intermediate results before passing them forward.

Example Chain

Step 1: Extract all dates and events from the document.
Step 2: Organize extracted events into a chronological timeline.
Step 3: Summarize the timeline into a narrative paragraph.

Key Takeaways

• Chain prompts when a single prompt would be too complex.
• Design each step's output to be the next step's input.
• Smaller, focused steps usually beat one massive prompt.