Prompt Refiner

Transform a vague or loose prompt into a precise, structured specification with concrete conditions and measurable outcomes.

What it does

Transform a vague or loose prompt into a precise, structured specification with concrete conditions and measurable outcomes.

Procedure

When this skill is activated, Chalie follows these steps:

  1. Use recall to surface any prior prompt work or specification preferences — then ask the user to share the prompt they want to improve and what they will use it for.
  2. Use write_file to analyse the original prompt and list its weaknesses: ambiguous terms, missing trigger conditions, unmeasurable outcomes, absent constraints, or undefined audience.
  3. Use write_file to apply EARS syntax transformation — rewrite each requirement using the appropriate pattern: Ubiquitous (“shall”), Event-driven (“When X, shall”), State-driven (“While X, shall”), Conditional (“If X, shall”), or Unwanted-behaviour (“shall prevent”).
  4. Use write_file to ground the rewritten requirements in a relevant domain framework (e.g. GTD, BJ Fogg, WCAG) — name the framework and explain how it informs each requirement.
  5. Use write_file to extract 2–3 concrete use-case examples with real data values rather than placeholders, and embed them in the specification.
  6. Use write_file to format the final output using the Role / Skills / Workflows / Examples / Output-format structure: one paragraph each.
  7. Present the refined prompt alongside a before/after comparison, highlighting the specific improvements made.
  8. Use write_file to save the refined specification if the user wants to keep it.

Version

v1 (curated)