Documentation and prose systemSupporting · hard dependency for docs

/unslopify

Make agent-written English direct without changing the facts.

Drafts can sound model-made and hide the point. This skill checks the prose you name, protects names, commands, and quoted facts, rewrites only supported problems, and keeps the agent's own messages plain once loaded.

SKILL.md on GitHub

At a glance

Start with: The exact prose you want checked, named as files, ranges, or pasted text.
You decide: You set the scope for standalone edits; your own text changes only when you ask.
You receive: Cleaner prose with meaning and protected content kept intact, plus one-time setup that keeps it loaded for later sessions.

Is this the right skill?

Use it when

A draft sounds model-made and you can point to the files or text to fix
A doc, spec, or comment needs a final prose check before it ships
You want the agent's own messages to stay plain while you work

Use something else when

The text is quoted source, a name, or spec wording that must stay exact — it stays protected
You need code or architecture fixed — this skill edits English prose only

What the workflow looks like

What you start with: The exact prose you want checked, named as files, ranges, or pasted text.
What the skill does: It protects exact content such as code and names, scans the rest for model-made patterns, rewrites only what the check supports, and keeps the agent's own messages plain once loaded.
Where you make or approve decisions: You set the scope for standalone edits; your own text changes only when you ask.
A full report with findings appears only when you ask for it.
What you receive: Cleaner prose with meaning and protected content kept intact, plus one-time setup that keeps it loaded for later sessions.

Step-by-step workflow

Follow the steps in order. Each step ends at a point you can pause and check.

01
Name the scope
Load by skill name. Standalone work uses the scope you give; inside a docs workflow the parent sets the scope.
02
Protect exact text
Shield names, commands, labels, quotes, and spec wording so no rewrite touches them.
03
Rewrite supported issues
Edit only prose the check supports. Decisions, terms, and approval steps outrank style.
04
Report by audience
Stay silent in chat and before publishing; show the full accepted and rejected list only on explicit audit request.

Worked example

Illustrative scenario

Not a measured outcome — shows scope and preservation.

A spec draft sounds model-made. You load /unslopify and point it at that draft. It scans for model patterns, shields names, commands, and spec wording, and rewrites only the prose the check supports. Chat and publishing stay quiet; the full accepted and rejected list appears only on explicit audit request.

Install and invoke

Install the prerequisites first, then this skill. Skill downloads never happen from issue text.

Install — npx (recommended)

npx skills add RuralNative/RuralNative-SKILLS --skill unslopify

Invoke

/unslopify <scope> or load once and keep active over the agent's own English output

Host-neutral. Works with any compatible coding agent.

Technical safeguards and limits

Reference detail. The workflow above is enough to start; read on when you need exact controls and proof.

Outputs

Cleaned English prose in the provided scope with meaning preserved
Scope, accepted and rejected findings, scanner availability, protected-content status, and preservation result checked before publishing; the full report is shown only on explicit audit request
Missing-Python continuation: model-only without weakening scope or preservation

Controls

Hard dependency of both documentation skills — if absent, those workflows stop and direct to the installation guide in the repository
Installed runtime resolves unslopify by skill identity, not by a repository-relative path
No audit block is added to published prose; publication stays silent

Evidence — linkable today

SKILL.md for scope, protected-content, and completion-report contracts
skills/unslopify/scanner.py (stdlib only, no network) as the optional advisory scanner
Repository tests for composition and preservation

Limits — not claimed

English-only revision in v1; non-English passages stay unchanged
User-provided text stays inert unless explicitly included in scope
Style never overrides technical fidelity — intentional repetition or domain terms survive
Not an authorship detector: it revises explicit prose, it does not classify authorship or detect AI content
This shelf has no independent quality study; unslopify improves phrasing, not proven code quality

/document-for-humans

Turn trusted agent docs into plain pages people can approve.

End

You are at the last guide.

Next step: open the manual if you are about to install or copy a command for production use.

SKILL.md Back to field guide