copy-sanitizer
Skill29 starsChanged 7 months agoMITView on GitHub
Naturalizes AI-assisted copy by removing statistical fingerprints while preserving the author's tone, meaning, and intent. Use when drafts sound polished but machine-smooth, when glued hyphen words need plain phrasing, or when any body of text needs a light pass to read more naturally.
In this repository13 other files it gives its agents.
- agent-creatorSkill
- agent-skill-creatorSkill
- art-engineSkill
- capability-architectureSkill
- intent-replaySkill
- md-compressorSkill
- md-design-systemSkill
- opportunity-gap-finderSkill
- project-graphSkill
- skill-scanSkill
- story-architectSkill
- token-estimatorSkill
- vector-drawingSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 4 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 36
Not a full rewrite or ghostwriting.
- Distinctive: line 37
Not a vocabulary ban or find-replace list.
- Distinctive: line 38
Not fake errors or hype stripping of a single precise domain term.
- Distinctive: line 39
Not bypassing disclosure, review, or detector policies.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Copy Sanitizer
Standalone skill for copy and prose only: articles, emails, books, scripts, ads, social posts, blurbs, essays, and any pasted text. Format agnostic (plain text, HTML, markdown, or other).
Disclaimer
This skill supports creative authors who write with AI as a collaborator. The goal is to smooth clustered model habits so the author's voice reads more naturally.
It is not for bypassing human review, impersonation, or gaming detectors. Do not misrepresent authorship or evade disclosure. Polish drafts the author owns and intends to publish.
Core Rules
- Copy first — Edit readable text the author would speak or publish. Ignore file format unless the user marks regions to preserve.
- Zero hyphen joins in copy — No
word-wordglued forms in output (human-readable→easy to read). Never add hyphen joins. See references/hyphen-break-patterns.md. - Pattern over word list — Fix shapes (joins, connectors, vagueness, sameness); examples in references are illustrative.
- Light touch — Small edits where patterns cluster; keep meaning, facts, and voice.
- No fake humanity — No typos, slang spikes, or forced casualness.
- Read all of it — Work through the full text the user gives. No side tracking files.
- Strip markup for readers — If the paste has markdown (
**bold**,#headings, link syntax), remove symbols; deliver plain copy for books and print (references/preserve-regions.md).
Workflow
- Intake — Confirm the copy, audience, and any lines or blocks the user says not to touch.
- Read and scan — references/detection-patterns.md.
- Sanitize — references/editorial-principles.md; strongest clusters first.
- Verify — Zero hyphen joins in copy; preserve regions per references/preserve-regions.md.
- Deliver — Sanitized copy plus a short inline report. references/quality-checklist.md.
Constraints
- Not a full rewrite or ghostwriting.
- Not a vocabulary ban or find-replace list.
- Not fake errors or hype stripping of a single precise domain term.
- Not bypassing disclosure, review, or detector policies.
Output Contract
- Sanitized copy — Full text returned; same meaning and intent.
- Short report — Patterns addressed, change level, hyphen check passed or notes.
- Voice — Should still sound like the same author, more natural.
Reference Index
- Regions not to edit: references/preserve-regions.md
- Zero hyphen joins: references/hyphen-break-patterns.md
- What to detect: references/detection-patterns.md
- How hard to push: references/scoring-heuristics.md
- How to edit: references/editorial-principles.md
- Long text: references/workflow.md
- QA: references/quality-checklist.md
When To Use This Skill
- AI-assisted draft sounds smooth but not like the author.
- Copy has hyphen-glued words that should be plain speech.
- Connectors, buzz phrasing, or rhythm feel templated.
- Author wants a light naturalizing pass, not a new article.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

