Critique story scenes with AI
Locate dialogue, setting and reaction problems in an authorised draft, with quoted evidence and prioritised revision directions rather than replacement prose.
Critique story scenes with AI when a passage reads correctly but still feels detached. The manuscript-prose-immersion-audit skill looks at relationships, dialogue, environmental presence and the emotional thread beneath a character’s actions. Findings should point to the actual passage rather than substitute a generic writing checklist.
The output gives revision directions, not rewritten prose. Your project’s audience and register guide the judgement, and a distinctive narrator should not be flattened into a rule. A sampled review must identify its coverage; it is no verdict on the whole book’s commercial prospects.
What it does
- Identify dialogue that explains already-known information
- Distinguish lived setting from an inventory of objects
- Locate long exchanges with little physical presence
- Check whether emotion runs through the action
- Protect passages that already achieve the intended effect
- Prioritise a small set of revision directions
Install and use
- Load the complete folder in a file-capable agent and provide authorised scenes. Supply the project profile if available; otherwise use a stated general standard. No extra runtime or account is supplied by this text package.
- Try: “Audit this station conversation between two fictional characters. Identify information-heavy dialogue, but do not rewrite my sentences.” State the audience and desired scope.
- Read each finding beside its quoted passage. Ask how the suggested direction serves your intention, and challenge a note that mistakes deliberate narration for a defect. Line numbers must come from the source.
- Choose the most useful changes yourself. Keep the same scope for a comparison pass; continuity, mystery logic and marketability belong to separate reviews.
FAQ
Source
Made by dinaf2026-web. Original skill: manuscript-prose-immersion-audit, published on GitHub under the MIT licence.
View on GitHub