HomeAboutProjectsSkillsAchievementsContact
🌙
Back to the radar

Claude Code AI Writing Humanizer Workflow

instagramreview
Original post

TL;DR

A workflow for Claude Code that uses curated rules and datasets (from the 'Humanizer' GitHub repo and the StoryScope research paper) to make AI-generated text read more like human writing. It is configured via a CLAUDE.md file so every human-facing document automatically runs a humanization pass. AI-generated text has recognizable stylistic patterns that reduce authenticity in marketing, emails, and content. This workflow systematically suppresses those patterns inside Claude Code for all human-facing documents.

What it actually is

  • What: A workflow for Claude Code that uses curated rules and datasets (from the 'Humanizer' GitHub repo and the StoryScope research paper) to make AI-generated text read more like human writing. It is configured via a CLAUDE.md file so every human-facing document automatically runs a humanization pass.
  • Who built it / maintained by: Jens Heitmann (workflow author); blader (Humanizer repo author)
  • Status: unknown
  • Why it matters: AI-generated text has recognizable stylistic patterns that reduce authenticity in marketing, emails, and content. This workflow systematically suppresses those patterns inside Claude Code for all human-facing documents.
  • How it compares to alternatives:
  • Grammarly
  • Quillbot
  • Undetectable.ai
  • GPTZero (detector)
  • Hemingway Editor
  • GitHub stars: 34,074 · License: MIT · Archived: no

Links

  • Repo: https://github.com/blader/humanizer

Kickstarter guide

Clone or reference the blader/humanizer repo on GitHub, which is based on Wikipedia's 'Signs of AI Writing' article. Incorporate rules from the StoryScope research paper on AI fiction idiosyncrasies. Merge both into a CLAUDE.md file at the root of your project with a rule requiring every human-facing document to run a humanization pass. Claude Code will then automatically apply these guidelines whenever it generates content.