omer-faruk-g

HumanSpeak — AI skill for Claude Code

AI community

HumanSpeak: Claude Code skill that strips AI buzzwords, robotic transitions and formula structure from English and Turkish text.

How to install HumanSpeak

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

What HumanSpeak does

HumanSpeak: Claude Code skill that strips AI buzzwords, robotic transitions and formula structure from English and Turkish text. Prompt-only, one SKILL.md.

Alternatives in AI

CLI 代理 API — English 中文一个为 CLI 提供 OpenAI/Gemini/Claude/Codex 兼容 API 接口的代理服务器 19k ★
DontFeedTheAI — Transparent anonymization proxy for AI-assisted pentesting 654 ★
Al Context.Create.Prompt — Generate or update context.md file documenting project structure, architecture, and key patterns for AI assist 109 ★

README

HumanSpeak

HumanSpeak is a text-rewriting skill for [Claude Code](https://code.claude.com). It is prompt-only: one `SKILL.md`, with no scripts and no hooks. It strips telltale AI vocabulary, robotic transitions and formulaic structure from English and Turkish text and rewrites it to read like a person wrote it.

Rules

Section	Rule	What the model does
1	Banned register	It removes generic Turkish openers ("Günümüzde...", "Şüphesiz ki..."), template connectors ("Ayrıca,", "Bunun yanı sıra,"), artificial emphasis ("Kritik bir rol oynamaktadır") and formula closings ("Sonuç olarak,"), and English buzzwords (delve, tapestry, pivotal, seamlessly...) and transitions (Furthermore, Moreover, In conclusion...).
2.1	No introductory fluff	The first sentence goes straight to the point, with no dictionary-style definition or generic setup.
2.2	Varied rhythm	It mixes very short sentences with longer ones instead of uniform 15-20 word sentences.
2.3	No formula conclusion	It drops closing paragraphs that only restate the text and ends on the last real point.
2.4	No symmetrical bullets	List items vary in length and form instead of all starting the same way.
2.5	Concrete over hype	Abstract praise becomes facts, numbers or direct statements.
3	Pipeline	Scan and scrub, restructure cadence, use direct phrasing, then a final check against common AI-writing patterns.
Output	Text only	It returns only the rewritten text, with no list of changes or commentary unless asked.

Install

git clone https://github.com/omer-faruk-g/HumanSpeak ~/.claude/skills/human-speak

Then restart Claude Code. The skill loads when you say humanize, insan gibi yaz or HumanSpeak, ask to de-AI a text, or type `/human-speak`.

Limits

The final audit is the model's own judgement. It does not run an AI detector, so nothing guarantees the text passes one.
It changes wording and structure, not facts. If