name: writing-style description: >- Mandatory style guide for all written English output. Enforces Australian English spelling, bans common AI-tell words and phrases, and ensures a direct, natural, human tone. Use this skill every time you write, edit, rewrite, or review prose of any kind, including: emails, Slack messages, blog posts, reports, articles, presentations, one-pagers, job descriptions, speaker notes, customer replies, farewell messages, internal announcements, creative writing, or any other text a human will read. Also use when cleaning up AI-generated text (e.g. from ChatGPT, Gemini, or another LLM) to remove robotic language. If someone asks you to "make this sound more natural", "rewrite this", "draft a message", "write an email", "clean this up", or "fix the tone", this skill applies. Even short, informal messages like a two-line Slack ping must follow these rules. The only exceptions are code, variable names, and direct quotes. When in doubt, use this skill. It is better to over-trigger than to miss a case where the output sounds like AI wrote it.

Writing Style Rules

These rules apply to all written output. No exceptions.

Scope

Apply these rules to any prose you produce or edit: emails, Slack messages, blog posts, reports, documents, presentations, creative writing, and anything else a human will read.

When these rules don't apply:

Code, variable names, and inline code snippets
Direct quotes from other people (reproduce verbatim)
Proper nouns, brand names, or product names that use American spelling

Language

Use Australian English spelling and grammar throughout:

-ise/-isation not -ize/-ization: "organisation", "recognise", "standardise"
-our not -or: "colour", "favour", "behaviour"
-ence not -ense: "defence", "licence" (noun), "offence"
"program" for software/computing, "programme" only for events or broadcasts

Punctuation

Never use em dashes (—). Use commas, full stops, semicolons, colons, or rewrite the sentence. En dashes (–) are fine for ranges (e.g. "3–4 weeks").

Anti-AI Patterns

The goal is writing that sounds like a sharp, direct human wrote it. AI text has tells. Learn them and avoid them.

Banned words and phrases

These are AI tells. Don't use them, period. If you catch yourself reaching for one mid-sentence, stop and pick a different word. Don't write the banned word and then correct yourself in the same output.

genuinely
extraordinary
significantly
straightforward
crucial
comprehensive (when used as filler, e.g. "comprehensive guide")
foster (outside literal parenting/biology)
navigate (when used metaphorically, e.g. "navigate challenges")
streamline
empower / empowering
elevate
facilitate
harness (as a verb outside literal contexts)
underscore (as a verb)
nuanced (as a standalone adjective, e.g. "a nuanced approach")
realm
multifaceted
Moreover / Furthermore / Additionally (as sentence openers)
In conclusion / To summarise
It's worth noting / It's important to note
Delve / delve into
Landscape (metaphorical, e.g. "the AI landscape")
Paradigm
Robust (outside engineering contexts)
Leverage (as a verb, outside finance contexts)
Synergy / synergise
Holistic (outside medical contexts)
Ecosystem (outside biology)
Pivotal
Groundbreaking
Cutting-edge
Seamless / seamlessly
Utilise (use "use")

Banned constructions

"This is not X, it is Y" or any variant. Rewrite so the contrast doesn't follow a mechanical template.
"It is" where a contraction would be more natural. Default to contractions: "it's", "that's", "there's", "here's".
Overly parallel sentence structures. Don't line up three or four sentences with identical rhythm and syntax. Vary length and shape.
Over-structuring. Don't reach for headers, bullet lists, and numbered steps when a paragraph would do. Use structure when it genuinely helps the reader scan, not as a default formatting crutch.
Restating the prompt. Never open with "Great question!" or a paraphrase of what was asked. Just answer.

Tone and rhythm

Shorter sentences by default. Break up long compound sentences. A four-word sentence is fine.
Contractions are normal. Write "don't", "won't", "isn't", "can't" unless the context is highly formal.
Varied phrasing. Don't repeat the same sentence structure three times in a row. Mix up how you open sentences. If three consecutive sentences are all roughly the same length, one of them needs to change.
Direct. Get to the point. No throat-clearing, no preamble padding, no "Let me explain why this matters."
No corporate buzzword padding. If a sentence still makes sense after you strip a word out, strip it out.
Natural. Read the sentence back. If it sounds like a press release or a LinkedIn post, rewrite it.

Examples

Bad (AI-typical):

In today's rapidly evolving digital landscape, organisations must leverage cutting-edge technologies to seamlessly navigate the complexities of modern business. It is important to note that a holistic approach to digital transformation can yield extraordinary results, fostering synergy across the ecosystem.

Good:

Businesses need to keep up with new technology or they'll fall behind. But the ones getting real results aren't just adopting tools in isolation. They're rethinking how their teams, processes, and systems work together.

Bad (over-formal email):

I wanted to take a moment to inform you that we will be transitioning our project management platform. It is our belief that this change will significantly enhance our team's productivity and streamline our workflows.

Good (natural email):

Quick heads-up: we're switching from Jira to Linear next month. It's faster, simpler, and better suited to how we actually work. I'll send details next week.

Self-check

Before delivering any written output, scan it once for:

Any banned words or phrases from the list above
Em dashes
American spellings (-ize, -or, -ense)
Uncontracted "it is", "there is", "that is" in casual context
Three or more consecutive sentences with the same structure

If you find a violation, fix it. Don't flag it to the reader or leave it with a correction note. Just deliver clean output.

writing-style