GaZmagik/iso-24495 / skills/iso-24495-4
iso-24495-4
VerifiedProvisional task skill for organisational plain language implementation (based on ISO/CD 24495-4, committee draft). Activates for plain language gap analysis, policy drafting, review workflow design, and organisational readiness for the future published standard. Does not activate for ordinary writing, rewriting, or reviewing of individual documents.
tests/fixtures/known-good/SOURCES.md
View on GitHub# Where these documents come from
This corpus measures how much noise the engine makes on writing that was not produced for it. Two kinds of document sit here, and the difference matters when reading the result.
## External documents
Written by other people, for real readers, before this project existed. These are the only documents here that can support a claim about false positives, because nothing about them was chosen to suit the rules.
| File | Source | Licence | Retrieved |
|---|---|---|---|
| `govuk-universal-credit.md` | https://www.gov.uk/universal-credit | Open Government Licence v3.0 | 2026-08-14 |
Contains public sector information licensed under the Open Government Licence v3.0. The text is reproduced unchanged; only the heading and paragraph breaks are ours, to make a valid markdown document.
**The adjudication came first.** Every expected result was written down before the auditor saw the text. That adjudication said 10 sentences, an average of 18.0 words, a longest sentence of 26 words, and no findings. The engine then agreed on every number. Had it disagreed, the disagreement would have been recorded as a false positive rather than explained away by editing the expectation.
That document sits at exactly the ten-sentence sample floor, so it is the only one here that exercises `sentence-average` at all.
## Written for this repository
The remaining files are ours, in registers the external set does not yet cover: public service instructions, consumer rights, technical guidance, warnings, names and measures, and international spelling.
They demonstrate the method and would catch gross drift. **They cannot support a false-positive rate**, for two reasons. They are short, and prose written by an agent working under this plugin's own output style is not neutral evidence about that style. Treat their zero as a smoke test, not as a measurement.
## Documents expected to fail
`../difficult/` holds the opposite control. Every document above is expected to pass, so on its own the corpus could not tell a working engine from a silent one. These are published documents that should produce findings.
| File | Source | Licence | Retrieved |
|---|---|---|---|
| `us-foia-statute.md` | 5 U.S. Code section 552, subsection (a)(1), subparagraphs (A), (C) and (D) | Public domain, as a United States federal statute | 2026-08-15 |
Adjudicated before the first run: four sentences of 11, 44, 31 and 28 words.
One `legalese` finding, two `sentence-length` findings, and no average, because four sentences sits below the floor. The engine agreed exactly.
**Re-adjudicated 2026-08-15.** A `complex-word` rule was added that day, and it reports `whereby` in subparagraph (A). The finding is real, so the expected result is now four rather than three. The rule set grew; the engine did not drift. A changed expectation is only honest when the reason is recorded.
A quoted or backticked span exempts a term only when the span names that term.
Longer quoted sentences remain subject to the same checks as other prose.
Subparagraph (B) and the introductory clause of paragraph (1) are omitted because they contain em dashes, which a repository test forbids in every markdown file. Nothing that remains is altered.
## Adding to this corpus
1. Choose the document before running the auditor, and record where it came from and under what licence.
2. Write down every expected finding, sentence count and average first.
3. Run the engine once, and record what it actually said.
4. If the two disagree, keep both. The disagreement is the result.
Do not require zero findings from an external source. Published prose has no obligation to satisfy this project's proxies, and a genuine finding is as informative as a clean pass.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| brainstormingobra/superpowersYou MUST use this before any creative work - creating features, building components, adding functionality, modifying behavior, or planning anything new, in software or out of it (a talk, a business, a renovation). | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |
| brand-guidelinesanthropics/skillsApplies Anthropic's official brand colors and typography to any sort of artifact that may benefit from having Anthropic's look-and-feel. Use it when brand colors or style guidelines, visual formatting, or company design standards apply. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| frontend-designanthropics/skillsGuidance for distinctive, intentional visual design when building new UI or reshaping an existing one. Helps with aesthetic direction, typography, and making choices that don't read as templated defaults. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| mcp-builderanthropics/skillsGuide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK). | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |