Use when every paragraph of a draft is fine on its own but the piece does not build to anything.
You are diagnosing the structure of a piece that reads well sentence by sentence and badly as a whole. You are not line editing.
Draft: {{DRAFT}}
The claim the piece is meant to land: {{CENTRAL_CLAIM}}
What the reader believes before they start: {{READER_STARTING_POINT}}
Output:
1. Section map table: Section | What it claims | What it assumes the reader already accepts | Does {{READER_STARTING_POINT}} or an earlier section supply that (yes/no) | Contribution to {{CENTRAL_CLAIM}} (load-bearing, supporting, decorative).
2. Breaks: every point where a section rests on something not yet established, quoted, with the earliest place it could be established instead.
3. Reordered outline with a one-line reason for each move, marking which moves are cosmetic and which change the argument.
4. Every section marked decorative: cut, demote to an aside, or keep, with the reason.
5. If nothing in {{DRAFT}} actually argues {{CENTRAL_CLAIM}}, say that first and stop before the reorder.
Rules:
- Work at section level. Do not rewrite prose.
- Do not invent connective material to paper over a break. Name the missing evidence instead.
- Judge contribution against {{CENTRAL_CLAIM}} only, not against how interesting a section is.
- No em dashes. British spelling.
Replace each placeholder with your own detail. The more specific you are, the less the model invents.
When should I run this rather than a line edit?
When every paragraph reads well and the piece still goes nowhere. A line edit improves the sentences and leaves the problem untouched. Run this first, because the reorder decides which sections survive and polishing beforehand spends the effort on material you are about to cut or demote.
What do I have to decide before running it?
The draft, the central claim you want it to land, and what the reader believes before they start. Write {{CENTRAL_CLAIM}} yourself rather than asking for it. If you cannot state it in a sentence, that is the finding, and the prompt is told to stop and tell you the same thing.
What does the structural map show?
A section map showing what each part claims, what it assumes, whether anything established that assumption, and whether it is load-bearing, supporting or decorative, then the breaks, a reordered outline, and a call on every decorative section. The breaks list is the useful bit, and it names missing evidence rather than hiding it.
What is the tempting fix that hides the problem?
Closing a break with a connecting sentence. The prompt refuses to invent connective material for a reason: the gap is missing evidence, and a smooth transition hides it from you as well as from the reader. Most long drafts need one absent paragraph rather than the rewrite they usually get.