THE TEXT LAB
Good words. Better tools.
A useful little workbench for writers, builders, and people who count tokens for fun.
01 / The numbers
Words use browser word segmentation when available; characters count Unicode code points. Tokens ≈ characters ÷ 4, a rough English heuristic—not a model tokenizer. Non-English text and code may differ widely.
02 / Readability
Flesch Reading Ease and Flesch–Kincaid grade use approximate English syllables and sentence boundaries. These describe surface complexity, not accuracy, accessibility, quality, or intelligence.
03 / Word frequency
Literal lowercased word counts. No stemming, semantic keyword extraction, or sentiment claims.
04 / Cleanup & conversion
Changes replace your text. Use Undo to restore the most recent change.
Case conversion is mechanical. Title and sentence casing can change acronyms and proper names; review before using.
05 / Prompt workbench
A small specification beats a giant incantation. Fill in the job and copy a reusable prompt.
This is a template utility, not an AI model or an automated prompt-quality score. Read the prompt guide.
06 / Compare two texts
Compare word overlap and count changes between two versions.
Jaccard overlap compares unique lowercased words. It is not plagiarism detection, semantic similarity, or an AI-text detector.