humanizer

name	humanizer
description	Mandatory pass for any text this agent writes into a project artifact: commit messages, PR descriptions, documentation, code comments, task files, chat replies saved to a file, anything shipped. Strips AI-writing tells. Required project-wide per AGENTS.md, not opt-in. Also callable directly for ad-hoc cleanup of a drafted passage.

humanizer

Apply humanizer_upstream.md in this directory (vendored from github.com/blader/humanizer, MIT license, see HUMANIZER_LICENSE). Read it in full and apply it.

Every check named in humanizer_upstream.md must be executed individually as its own action, with its result shown. A single overall reading of the text does not satisfy this requirement. The skill is not complete until every one of its named checks has been executed this way.

Do not hand-edit humanizer_upstream.md. It is a wholesale vendor copy; project-specific voice rules belong in a separate file a task-specific skill can add, not in this one. Refresh by re-fetching entirely:

curl -s https://raw.githubusercontent.com/blader/humanizer/main/SKILL.md \
  -o developer/skills/humanizer/humanizer_upstream.md