Diagnostic Analysis Model
How it works
When you submit writing to the Diagnostic:
- Your text is sent to a single AI model (currently Claude Sonnet)
- The model reads the full submission
- It evaluates the writing across defined craft dimensions
- Results are returned as a structured analysis with per-dimension scores and specific feedback
What it evaluates
The Diagnostic assesses multiple dimensions of writing craft:
Voice — Is the narrative voice consistent? Does it match the content's intent? Is it distinctive or generic?
Structure — Does the piece have clear organization? Do sections/paragraphs flow logically? Is there a coherent arc?
Pacing — Does the writing move at an appropriate speed? Are there sections that drag or rush? Does scene length match scene importance?
Clarity — Is the writing easy to follow? Are sentences structured for comprehension? Is jargon used appropriately?
Dialogue (if applicable) — Does dialogue sound natural? Does it advance the story or just fill space? Are characters distinguishable through speech patterns?
Description — Is sensory detail present and effective? Is the writing showing or telling? Is description proportionate to its narrative importance?
Mechanics — Grammar, punctuation, and sentence construction. The least important dimension — mechanics matter, but they're the easiest to fix and the least informative about craft.
Strengths and limitations
The Diagnostic is fast, consistent, and good at identifying broad patterns in your writing. It excels at catching structural issues, voice inconsistencies, and pacing problems.
Its limitation is that it's a single model's perspective. A Diagnostic that says your pacing is slow reflects one analytical lens. The Jury provides multiple perspectives for higher-confidence assessment.
Best used for
The Diagnostic works best during the drafting process — submit a chapter after your first revision, review the feedback, revise, and submit again. The rapid turnaround supports iterative improvement.