humanizer

An agent skill by blader, from blader/humanizer. Tags: copy-editing, debugging, documents, generative.

What it does

Rewrite AI-sounding text so it reads like the writer without changing what it says. Use when editing or reviewing prose for AI tells: not-X-but-Y contrasts, one-line closers, staged openers, forced triads, dashes everywhere, inflated claims, sales language, stock AI words, bold labels, or filler. Based on Wikipedia's "Signs of AI writing."

Install

With the skills CLI, which installs into Claude Code, Codex, Cursor and other agents:

npx skills add blader/humanizer --skill humanizer

Or copy the skill folder into Claude Code's skills directory by hand (~/.claude/skills for every project, or .claude/skills inside one):

git clone --depth 1 https://github.com/blader/humanizer
cp -r humanizer ~/.claude/skills/humanizer

Safety box score

Grade C, moderate risk: 28 of 100 risk points, rated 2026-09-22. The grade rates what the skill and its scripts can reach on the machine of whoever installs it, across eight categories from shell execution to secrets access. How the score works.

This is a text-rewriting skill, not a code-execution or data-collection tool. Its main blast radius is that it can rewrite and write text files back to paths the user names, and its install/validation path uses `npx` and `npm install --global`, which pull packages from the registry.

Category levels, from 0 (none) to 3 (high)
Category	Level	Why
Code & shell execution	1 (low)	It ships a local validation script (`scripts/validate-package.py`) and the docs/workflow run registry commands like `npx skills add ...` and `claude plugin validate .`, but the skill itself is just Markdown prompt text.
Network access	1 (low)	The README and workflow use `npx` and `npm install --global`, which download packages from npm during install/validation, but the skill does not otherwise send data to remote hosts.
Filesystem reach	2 (moderate)	File mode writes rewritten prose back to the path the user names: `"When the user names a file, run the full process but write only the final text to the file."` That is bounded to prose, but not to a single directory.
Secrets & credentials	0 (none)	There is no instruction to read `.env` files, keychains, browser sessions, or other credential stores; the mention of tokens is only in examples about text to edit.
Privilege & persistence	2 (moderate)	The install docs and CI use global package installs (`--global` and `npm install --global`), which change user-level package state even though they do not require sudo.
Instruction hijack surface	0 (none)	The skill tells the agent to treat the text as material to edit, not as instructions to follow, and it does not ask it to ignore confirmations or fetch outside instructions.
Irreversible actions	1 (low)	In file mode it automatically writes the final rewrite to disk, but it does not send messages, publish, pay, book, deploy, or delete.
Transparency	0 (none)	All shipped files are readable Markdown, YAML, JSON, or Python; there are no binaries, bundled installers, or decoded-at-runtime blobs.

Before installing

Read the file-mode section in `SKILL.md` and confirm you are okay with it writing back to user-named paths.
If you plan to use the install/validation commands, try them in a disposable project or account first because they use `npx` and `npm install --global`.
Prefer the non-global install path if you do not want user-level package changes.
Check `.github/workflows/validate.yml` if you want to mirror CI locally; it installs `@anthropic-ai/claude-code` globally.

Rated 2026-09-22 by an automated review of 8 files from the skill's repository.

Source

Repository: blader/humanizer (all skills from this repository)
Path: SKILL.md
Branch: main
Updated: 2026-09-19

grammar-check — Identify grammar, logical, and flow errors in text and suggest targeted fixes without rewriting the entire text.
writing-guidelines — Review docs/prose for Writing Guidelines compliance.
enhance-prompt — Transforms vague UI ideas into polished, Stitch-optimized prompts.
google-docs — Author and edit Google Docs with the google-docs tools - structured writing, character and paragraph styling, lists, tables, images, headers/footers.
archify-review — Review Archify issues, PRs, or code through value, cost, and impact to support evidence-based maintenance decisions.
book-sft-pipeline — This skill should be used for book-to-SFT pipelines: ePub extraction, literary segmentation, author-voice dataset construction, style-transfer training.