Перейти к содержимому
TakoAPI
Назад к навыкам

nlp-toolkit

от raghulpasupathi

Search & Research
459Число загрузок по данным clawskills.sh
1Просмотры
0 лайковВойдите, чтобы поставить лайк навыку
—Ещё нет оценок
Ваша оценка

Кратко

Advanced NLP with perplexity scoring, burstiness analysis, and entropy calculation.

Описание

Что делает этот навык

Analyzes text using statistical NLP measures to estimate whether it was written by a human or generated by AI. Computes perplexity, burstiness, and Shannon entropy scores, then combines them into a confidence rating.

Runs entirely client-side with no external API calls, so text is never sent to a third-party detection service.

Когда его использовать

  • Screening submitted essays for AI authorship
  • Flagging AI-generated support tickets before human review
  • Auditing content pipelines for model-generated text
  • Comparing writing samples before and after AI tool adoption
  • Spot-checking blog drafts for undisclosed AI use

Пример сценария

ВВОД

User asks: Is this 200-word paragraph AI-generated?

АГЕНТ
  1. 1

    Pass the text to analyzeText with default thresholds

  2. 2

    Calculate perplexity from unique-to-total word ratio

  3. 3

    Measure burstiness as coefficient of variation across sentence lengths

  4. 4

    Compute Shannon entropy over character frequency distribution

  5. 5

    Combine three scores into a single confidence percentage

ВЫВОД

{ isAI: true, confidence: 78, metrics: { perplexity: 32.4, burstiness: 0.21, entropy: 4.1 }, explanation: 'Low perplexity and uniform burstiness suggest AI generation' }

Установка

clawhub install raghulpasupathi/nlp-toolkit

Выполните команду CLI или вставьте промпт в Claude Code, Cursor, Codex — любой кодинг-агент, которым пользуетесь.

Ссылки