Which one would you keep reading?
Same brief, two anonymous drafts from the same model. One had a writing skill loaded. Tap the one you prefer.
$ npx skills add in-c0/writing-skill --skill storyscope -g
Installs StoryScope for Claude Code, Codex, Cursor and other agents. Other ways
Results
Six briefs, four setups, one writing model. Two AI editors scored every draft blind. Then the same drafts were scored again with a prompt that never mentions AI.
By genre, classifier and lint
All 24 drafts
Unedited outputs. Marks show surface patterns such as em dashes, triads and stock words; they are hints, not proof.
Judge three rounds
Four anonymous drafts per round. Pick the best and the worst; your picks can be added to the human column of the results.
Use it
In an agent
$ npx skills add in-c0/writing-skill --skill storyscope -gSwap storyscope for writing-skill-storyscope or writing-skill. Drop -g to install into one project.
In the Claude app
Download storyscope.zip and upload it under Skills in settings. Also available: writing-skill-storyscope.zip, writing-skill.zip.
In any chat
Paste the prompt, then add your task after the last line.