Learn

Structure Reusable Prompts

Write prompts that are explicit, maintainable, and aligned with schema-first Cuddler workflows.

Prompt anatomy

What a reusable prompt needs

Reusable prompts work when they carry the task, the source facts, the governing context, and the output constraints explicitly.

JobWhat the prompt should accomplish

State the task in terms another author could reuse without reinterpretation.

Source factsWhat the assistant should rely on

Name the inputs or evidence the prompt is allowed to use.

ContractWhich schema or artifact family applies

Keep the prompt anchored to the right versioned family and terminology.

Output constraintsWhat success should look like

Separate required content, optional content, and forbidden shortcuts.

If the rules belong in the schema, move them there instead of hiding them in the prompt.

Reusable prompts work best when they describe the task, the expected input shape, and the output constraints without forcing the reader to guess.

  1. Start with the job the prompt should accomplish.
  2. Name the source facts the assistant should use.
  3. State which schema or artifact family the prompt belongs to.
  4. Keep the instructions concrete enough that another author could reuse them without reinterpreting them.
  5. Match the prompt to the versioned domain or artifact family it supports.

Good Prompt Traits

  • The prompt says what success looks like.
  • The prompt separates required content from optional content.
  • The prompt uses the same terminology as the public standards pages.
  • The prompt does not smuggle in rules that belong in the schema or specification.