Wayne AI Tell Scanner

Wayne-mode flags · in-browser perplexity detector · voice-rewrite briefs for Claude.

Wayne-mode audit

Cut the filler. Break the rhythm. Put some blood back in the sentence.

This pass scores cadence, repetition, hedging, abstraction, passive drift, and soft openers. It also runs distilGPT-2 in your browser to measure how predictable the prose is. For voice-level rewrites — the kind where "the damage" becomes "the doorway you stop in without knowing why" — it builds a Claude-ready brief you can paste straight into your real writing tool.

Mechanical rewrites are shallow.The auto-rewrites cut hedges, split long sentences, and swap abstract nouns for plain ones. They don't write in your voice — only Claude or you can do that.
Voice-rewrite brief, ready to paste.Each scan produces a Claude-ready prompt invoking your voice profile (Stein show-don't-tell, Ackerman embodied emotion). Copy it. Paste into Claude. Get real rewrites.
Translation, not just numbers.Every score has a plain-English reading + a concrete "do this." No mystery metrics.

Draft input

Paste text or upload a text-based file.

No file loaded
Text files, PDFs, and Word `.docx` documents. PDF/DOCX parsers (PDF.js, mammoth.js) load on first use and cache locally. Scanned/image PDFs are not extracted — no OCR.

Perplexity from a small GPT-2 model is a triage signal, not proof. Some humans write predictably; some LLM output is high-perplexity. Treat it as a second opinion on your Wayne flags.

Risk readout

Read the weak joints first.

No scan yet
AI-sheen score (Wayne)
0
Perplexity (distilGPT-2)
—
Runs on first Scan (~80 MB one-time download)
Top issue
None
Reading
Paste text and run a scan.

Flags

What the scores mean

Run a scan to see plain-English readings and concrete next steps.

Shallow mechanical rewrites

Top 3 lines

Rule-based cleanup only — cuts hedges, swaps abstract nouns, splits long sentences. Treat as a first pass, not as your voice.

Original lines (with triggers)

Mechanical rewrites

Wayne-voice rewrite brief (paste into Claude)

For real voice-level rewrites — concrete sensory detail, embodied emotion — paste this whole brief into Claude (or another instruction-tuned LLM). The mechanical rewrites above can't do this; Claude can.

Brief will appear after you scan a draft with at least one flagged line.

Rewrite marching orders (per category)

No report generated yet.