name: iso-24495 description: "Use when text must be easy for its intended reader to find, understand, and act on — reports, emails, documentation, UI text, announcements, explanations, AI output that reads as dense or bureaucratic. Applies ISO 24495-1 plain language principles with dedicated technique layers for English and Traditional Chinese. Triggers: plain language rewrite, apply ISO 24495, make this clearer for readers, de-jargon this, 淺白改寫, 白話改寫, 去公文腔, 讓讀者一次看懂. Not for creative or literary writing." version: 0.1.0

ISO 24495-1 Plain Language (English + Traditional Chinese)

ISO 24495-1:2023 is the international standard for plain language. It defines plain language by reader outcome, not by word lists: a document is plain when its intended readers can get what they need (relevant), find it (findable), understand it on first reading (understandable), and act on it (usable).

This skill rewrites text so it meets those four outcomes. The principles are language-independent — ISO states the standard "applies to most, if not all, written languages" — so this skill ships one shared principle layer and two language technique layers:

• English → apply references/english-techniques.md
• Traditional Chinese（繁體中文） → apply references/chinese-techniques.md

The Chinese layer is not a translation of the English layer. It is a separate technique set for problems Chinese text actually has (歐化長句、公文腔、 成語堆疊、中英夾雜、被字句濫用). As of this skill's release, no national standards body has published a Chinese adaptation of ISO 24495-1; this layer is an original implementation of the principles for Chinese.

When to Use This Skill

• A reader-facing text (report, email, doc page, UI string, announcement, policy, explanation) reads as dense, bureaucratic, hedged, or jargon-heavy.
• AI-generated output is technically correct but hard to read.
• Text must work for non-specialist readers, non-native readers, or busy readers who will scan rather than study.
• You want a before/after with the violated principle named per change — ask for it; the default output is the rewritten text alone.

Not for creative or literary writing, and not for agent-facing controlled language — for text a machine must parse without ambiguity, use a controlled language standard such as ASD-STE100 instead (see Scope Boundary below).

Procedure

1. Identify the reader and purpose (Principle 1 requires it). If the user named an audience, use it. Otherwise assume a busy, intelligent, non-specialist adult reader and state that assumption in one line only when it changes the rewrite materially.
2. Detect the text language. English → English techniques file. Chinese → Chinese techniques file. Mixed text → the dominant language's file, and handle embedded foreign terms per that file's rules. Other languages → apply the four principles directly and say the technique layer for that language does not exist yet.
3. Read the whole text for meaning before touching a sentence.
4. Rewrite against the four principles, using the language file's techniques. Never drop a fact, condition, number, or scope qualifier. If plainness would cost precision, keep the precision and flag the trade-off in one line.
5. Run the language file's self-check list before returning.

Output Format

Return the rewritten text alone — no preamble, no principle lecture, no change summary. Append a one-line Kept as-is: note only when something was deliberately left unsimplified for precision.

When the user asks to "show the diff", "explain the changes", or "which principles", return a before/after table instead, one row per change, naming the principle (relevant / findable / understandable / usable) and the technique applied.

The Four Principles (working definitions)

Full interpretation, sources, and copyright boundaries: references/principles.md.

Scope Boundary: Plain Language vs Controlled Language

ISO 24495-1 optimizes text for human readers and permits full natural language. Controlled-language standards such as ASD-STE100 optimize for zero-ambiguity parsing (originally by maintenance technicians, now also by AI agents) and restrict vocabulary and grammar hard. If the reader is a machine or the cost of one misread word is high, use a controlled-language skill. If the reader is a person who needs to find, understand, and act, use this one.

Copyright Note

This is an unofficial project, not affiliated with or endorsed by ISO. The full ISO 24495-1 text is a licensed document and is not reproduced here. This skill is built from the standard's publicly documented principles and public plain-language practice; every quantitative rule in the technique files is this project's own working proxy, not a clause of the standard. For certified conformance work, buy and consult the standard itself.

ISO 24495 Skill — Plain Language for English and Traditional Chinese

A Claude Code skill that rewrites text into plain language per ISO 24495-1:2023 — the international standard that defines plain language by reader outcome: the intended readers can find what they need, understand it on first reading, and use it.

Two things make this skill different from a one-line "write plainly" prompt:

1. It implements the standard's actual structure — reader-first content selection (relevant), document structure (findable), wording (understandable), and actionability (usable) — not just shorter sentences.
2. It ships a real Traditional Chinese technique layer（繁體中文層）. ISO 24495-1 is explicitly language-independent, but as of this release the official translation list contains no East Asian language at all. The Chinese layer here is an original implementation of the four principles for Chinese — targeting the problems Chinese text actually has（歐化長句、公文腔、成語堆疊、中英夾雜、 被字句濫用）, not a translation of the English rules.

Before / After

More pairs, each naming the principle applied: examples/before-after-en.md · examples/before-after-zh.md.

What This Skill Does

1. Identifies the intended reader and purpose — the standard's first principle requires it, and every cut/keep decision follows from it.
2. Detects the text language and loads the matching technique layer: references/english-techniques.md or references/chinese-techniques.md.
3. Rewrites against the four principles (relevant → findable → understandable → usable) without dropping any fact, number, condition, or honest hedge. Where plainness would cost precision, it keeps the precision and flags the trade-off.
4. Returns the rewritten text alone — no preamble, no lecture. Ask for "show the diff" and it returns a before/after table naming the principle behind each change instead.

Architecture: one shared layer, per-language technique layers

SKILL.md                          shared layer (English): principles, procedure, routing
references/principles.md          the four principles, interpretation, sources
references/english-techniques.md  English techniques (in English)
references/chinese-techniques.md  中文手法（以中文撰寫）

The shared layer is language-independent, mirroring the standard itself. Technique layers are written in their own language, because plain language techniques do not translate: English fights nominalization and passive voice; Chinese fights Europeanized long sentences and bureaucratic register. Adding another language = adding one technique file. Contributions welcome.

Installation

git clone https://github.com/danyuchn/iso-24495-skill ~/.claude/skills/iso-24495

Usage

Rewrite this in plain language
Apply ISO 24495 to this announcement
把這段改成淺白中文
幫我去掉這封信的公文腔

You get the rewritten text back and nothing else. Add "show the diff" / 「列出改了什麼」 to see the per-change principle table.

Scope

Built for: reader-facing text — reports, emails, documentation, UI strings, announcements, policies, explanations, and AI output that reads as dense or bureaucratic. Works on English and Traditional Chinese; the principles apply to other languages but no technique layer exists for them yet.

Not built for: creative or literary writing (voice is the point there), and not for agent-facing text where a machine must parse without ambiguity — that is controlled-language territory; use the companion skill asd-ste100-skill instead. Rule of thumb: reader is a person → this skill; reader is a machine or a maintenance manual → STE.

Relationship to the ISO Standard

This is an unofficial project, not affiliated with, endorsed by, or approved by ISO. The full ISO 24495-1 text is a licensed document and is not reproduced here. The skill is built from the standard's publicly documented principle framework (ISO's published abstract and scope, and the International Plain Language Federation's freely published materials). Every quantitative rule in the technique files is this project's own working proxy, not a clause of the standard. For certified conformance, buy and consult the standard.

Sources

• ISO 24495-1:2023 — Plain language, Part 1
• International Plain Language Federation — ISO standard
• IPLF — national adoptions and translations
• AccessibleEU — ISO 24495-1 summary
• Companion controlled-language skill: asd-ste100-skill

License

MIT — see LICENSE. The license covers this repository's original text; it grants no rights to any ISO publication.