artsy Plain English
Vendored into artsy/duchamp from a local skill, MIT licensed, see LICENSE in this directory. The frontmatter
descriptionis the only change: it triggers on writing a PR review rather than on a user request, since no user speaks during an automated review.
Strip prose of two layers of bad habits:
- Classical bloat — passive voice, abstract subjects, Latinate padding, dying metaphors. The Orwell/Gowers tradition.
- AI tics — the recognisable LLM dialect: em-dash overuse, banned vocabulary, preamble openers, summary closers, reflex rule-of-three lists, false balance, sycophancy.
Modes
| Mode | When it fires | What to render |
|---|---|---|
| Audit | User pastes prose and asks for critique | For each flagged sentence: original → one-line flag (e.g. passive without agent, abstract subject, banned word: leverage) → suggested rewrite. Don't lecture. Don't restate the rules. |
| Rewrite | User asks for a rewrite, OR this skill is invoked as a self-audit before delivering long-form output | Cleaned prose first. If the user asked for explanation, follow with 2–3 bullets of what changed. Bullets, not paragraphs. |
Core rules
Orwell/Gowers — apply first
- Cut every word that adds nothing. If removing it leaves the meaning intact, it goes.
- Active voice over passive. If you can name the agent, name it.
- Concrete nouns over abstract. Never start a sentence with an abstract noun ("the realisation of expectations…") if a person or thing can do the work.
- Short word over long. Saxon over Latinate. Use over utilise. Help over facilitate. Before over prior to.
- Single word over circumlocution. Because over due to the fact that.
- No dying metaphors. If the phrase has been printed a thousand times (toe the line, Achilles' heel, at the end of the day), kill it.
- Break any rule rather than write something barbarous. Clarity wins.
AI detox — apply second
- Banned vocabulary. See
REFERENCE.md. Hard list including delve, tapestry, navigate, leverage, landscape, ecosystem, realm, multifaceted, foster, underscore, robust, comprehensive, nuanced, paramount, crucial, holistic, pivotal. Substitute or delete. - Em-dash budget. Maximum one em-dash per ~200 words. Default to commas, full stops, or new sentences. Em-dash overuse is the loudest AI tell.
- No preamble. Don't open with "That's a great question," "Certainly," "I'd be happy to," or framing of the upcoming answer. Start with the answer.
- No summary closer. Don't end with "In conclusion," "To sum up," "I hope this helps," or a paragraph that restates what was just said.
- No false balance. When one side genuinely outweighs the other, say so. No reflex "on the other hand."
- No reflex rule-of-three. If the content has two points or four, use two or four. Don't pad to three.
- Vary sentence length. AI prose clusters around 18–22 words. Good prose ranges from 4 to 40. Mix short punchy sentences with longer ones.
- No sycophancy. Don't validate the user's framing before answering. Don't say "great point."
- Cut hedge stacks. Genuinely, honestly, it's worth noting, I find that, arguably — one hedge per claim, max. Stacked hedges erase the claim.
Audit checklist
The mechanical application of the Core rules above. Used by both modes — audit reports flags, rewrite acts on them.
Two steps. Don't collapse them.
- Flag first. Walk every sentence against the Core rules. Do not pre-filter for voice, rhythm, or "the user's style." If a rule fires, mark it.
- Override second. For each flag, decide whether rule 7 (break the rule rather than be barbarous) applies. Name the reason in one word: rhythm, emphasis, picture, idiom, joke. If no reason can be named, take the fix.
Walk the checklist:
For each sentence, ask in order:
- Active or passive? If passive without reason → fix.
- Any word that could be cut without changing meaning? → cut it.
- Verbal false limb? (make contact with, exhibit a tendency to, give consideration to) → replace with a single verb.
- Abstract noun as subject? → make it concrete or use a person.
- Banned LLM vocabulary? Latinate where Saxon fits? → substitute.
For the whole text:
- Em-dash count vs word count → trim if over budget.
- Preamble? Summary closer? → delete.
- Sentence length variation → flag if uniform.
- Reflex three-part lists → check the count is real, not padded.
When NOT to apply
- Direct quotes from human sources → leave them.
- Code, technical specifications, legal text → don't "tighten" jargon that's load-bearing.
- The whole text is in deliberate dialect (e.g. AAVE, Scots, period pastiche) → ask before changing. Per-sentence voice judgments belong in the override step above, not here.
- Fiction and creative writing → these rules are for non-fiction. Fiction has its own logic.
Reference
See REFERENCE.md for the full banned-word substitution table and before/after examples. Load it when running an audit or rewrite — that's where the substitution table lives.