Walid-Idrissi-Labs

SciHumanizer — AI skill for Claude Code

AI community

A Claude skill that writes and rewrites English and French prose without the measurable habits of AI text, then checks the result with scripts: style bands calibrated on human writing, a fact ledger.

How to install SciHumanizer

This entry records only its repository, not the path inside it, so there is no exact command to give. Open Walid-Idrissi-Labs/SciHumanizer and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What SciHumanizer does

A Claude skill that writes and rewrites English and French prose without the measurable habits of AI text, then checks the result with scripts: style bands calibrated on human writing, a fact ledger, and a length check that refuses padding.

Alternatives in AI

  • Open Multi Agent — TypeScript multi-agent orchestration engine — one runTeam() call from goal to result 5.8k ★
  • Om Humanize — Voice-calibrated editing — makes Claude-drafted text sound like you wrote it, not like AI wrote it 4.6k ★
  • Surface — Show ranked attack surface for a target from its recon manifest + hunt memory 3.8k ★

README

SciHumanizer

English · [Français](README.fr.md)

MIT licence Python 3.8 or later English and French

SciHumanizer is a skill for Claude that writes and rewrites English and French prose so that it no longer reads as machine-generated, then verifies its own work with scripts. Although we built it for scientific and academic writing (paper sections, abstracts, theses, reports), it is also calibrated for essays, articles and opinion pieces.

When a text "sounds like AI", the usual treatment is a list of words to avoid. The research behind this project shows why such a list is not enough: in one study, a detector still reached 99.5 F1 on text generated with style prompts such as "write in the style of a high-school student". What gives machine prose away is mostly statistical. It sits in the grammar, in sentence length and in the proportion of names and numbers. Nobody can see it by rereading a draft, so the skill measures it. For every draft, an analyzer makes the comparison with human writing in the same language, in the same register and of the same length. A fact ledger then verifies that no number, citation, negation or hedge has changed, while a length check refuses to reach a word count with filler.

A rewrite, before and after

On the left is a sentence from an introduction that a model drafted from an abstract (the study is invented; it serves as a test case). On the right, the rewrite keeps every fact and every technical term. Only the grammar changes: one long sentence becomes three, with the verbs and articles that the draft had squeezed out.

Before After
Age-related declines in thermoregulatory capacity, a higher prevalence of chronic cardiovascular, respiratory, and renal condit