@domandigital/craft 0.7.0 → 0.8.0

package/CHANGELOG.md

@@ -1,5 +1,16 @@
 # @domandigital/craft
 
+## 0.8.0
+
+### Minor Changes
+
+- 2b6f823: The house copy standard moves into craft. `COPY.md` ships in the package, `craft copy --gate` applies the house policy (the blocking tier fails the run), and `craft tells list --json` publishes each copy tell's house tier so `copy-check` reads the policy instead of keeping its own list. Catalogue 2026.09.4 adds the density tier (`phrase-density`, `aphorism-density`, `contraction-scarcity`, `sentence-rhythm`, `repeated-sentence`, `heading-shape`, `heading-echo`) and four review tells from research (`ing-tail`, `vague-attribution`, `closing-summary`, `false-range`), all warn. The emoji tell now catches sparkles, the green tick box and other emoji in the older symbol blocks.
+
+  Words inside a URL or a Markdown link target are no longer read as copy, so a slug such as `/seamless-booking` no longer blocks.
+
+- 7765dda: Catalogue 2026.09.6, tuned against the Doman Digital site's live Sanity content. `chatbot-residue` reads a run of ChatGPT citation tokens as one finding, through the invisible private-use characters ChatGPT wraps them in. `placeholder` skips a placeholder inside quotation marks, which is a template being taught. `vague-attribution` skips a claim that names its source or carries a footnote or link. `inline-label-list` fires only when the text after the label is six words or fewer, so a definition list passes.
+- ef5b85b: Catalogue 2026.09.5. Four more review tells: `chatbot-residue` (pasted chat replies, chat citation tokens, links tracked `utm_source=chatgpt.com`), `placeholder` (`[Insert ...]`, lorem ipsum, `TODO_PLACEHOLDER`, including a placeholder that is a whole element), `question-reveal` and `inline-label-list`. New command `craft copy compare <before> <after>`: the preservation gate, listing the protected facts (prices, numbers, dates, times, phones, emails, links, postcodes, names) a rewrite lost or added, exit 1 if any. Exported as `compareFacts` and `protectedFacts`.
+
 ## 0.7.0
 
 ### Minor Changes

package/CHARACTER.md

@@ -279,14 +279,16 @@ navigation, the order of sections, where the call to action sits.
 
 ## The house copy rules live here
 
-Since catalogue `2026.09.2` the rule lists of claude-kit's
-`house-style/copy-rules.md` are carried by tells in this catalogue:
-`ai-phrase`, `plainer-word`, `plain-english`, `buzzword`,
-`negative-reassurance`, `em-dash`, `emoji`, `not-just-but`, `no-x-no-y` and
-`no-x-badge` are its blocking tier; `review-phrase` and `vague-word` its review
-tier. craft ships all of them as `warn`. `copy-check` in claude-kit is a thin
-wrapper over `craft copy --json` and decides which tells fail a commit, so the
-lists exist once, here, with a test per entry.
+The house copy standard is [COPY.md](./COPY.md), in this package. Its
+blocking tier is carried by ten tells here: `ai-phrase`, `plainer-word`,
+`plain-english`, `buzzword`, `negative-reassurance`, `em-dash`, `emoji`,
+`not-just-but`, `no-x-no-y` and `no-x-badge`. craft ships every tell as
+`warn`; the house policy in `src/character/house.ts` says which fail a commit,
+and `craft copy --gate` applies it. `copy-check` in claude-kit is a thin
+wrapper that reads the same policy from `craft tells list --json`, so the
+lists and the policy exist once, here, with a test per entry. Since catalogue
+`2026.09.4` the density tier and four research tells are here too, and since
+`2026.09.5` four more, including `chatbot-residue` and `placeholder`. All review.
 
 The swap was checked against the old checker on 2,570 files across the seven
 client repos (2026-09-23). The old checker found 193 issues, craft finds every
@@ -532,7 +534,7 @@ Generated from the package. Run `pnpm --filter @domandigital/craft run docs`
 after changing an entry; a test fails until you do.
 
 <!-- craft:catalogue:start -->
-Catalogue version `2026.09.3`, 39 tells.
+Catalogue version `2026.09.6`, 54 tells.
 
 | Id | Gen | Surface | Severity | Tell | Why it is a default |
 | --- | --- | --- | --- | --- | --- |
@@ -575,4 +577,19 @@ Catalogue version `2026.09.3`, 39 tells.
 | `no-x-badge` | 2 | copy | warn | 'No X' badge | 'No catch.', 'NO JARGON GUIDE': a short standalone 'No X' reads as a slapped-on kicker, and worse when the same one is reused across pieces. |
 | `staccato-triplet` | 2 | copy | warn | Staccato triplet | Three fragments in a row ('Fast. Friendly. Local.' or 'No fuss. No jargon. Just results.') is the second wave's favourite rhythm. |
 | `where-x-meets-y` | 2 | copy | warn | 'Where X meets Y' | 'Where luxury meets comfort' is a tagline shape that fits every business and so describes none. |
+| `chatbot-residue` | 1 | copy | warn | Chatbot residue | 'Certainly! Here's a revised version', 'as an AI language model', a citeturn0search0 token, a link tracked utm_source=chatgpt.com: text that exists only because a chat reply was pasted whole. This is evidence, not style. |
+| `placeholder` | 1 | copy | warn | Unfilled placeholder | '[Insert testimonial]', '[Your Name]', lorem ipsum, TODO_PLACEHOLDER: scaffolding a reader must never see, and the clearest sign a page shipped before anyone read it. |
+| `question-reveal` | 2 | copy | warn | Staged reveal | 'The result? Faster growth.', 'What does this mean for you? It means...': the writer asks the reader's question for them and answers it. An infomercial hook, and a favourite of generated copy. |
+| `inline-label-list` | 2 | copy | warn | Bold-label bullets | '**Speed:** Faster pages' three times in a row: a short label restated in a few words, the list shape generated copy defaults to. A definition list, with a real explanation after each label, is not this. |
+| `ing-tail` | 2 | copy | warn | Empty -ing tail | '..., ensuring peace of mind', '..., highlighting our commitment': a participle tacked on the end that claims significance and names no mechanism. One of the most common shapes in the Wikipedia guide to AI writing. |
+| `vague-attribution` | 1 | copy | warn | Unnamed source | 'Studies show', 'experts agree', 'it is widely known': a claim credited to nobody. A reader cannot check it, and the house proof rule forbids a claim nobody can check. |
+| `closing-summary` | 1 | copy | warn | Closing summary | 'Overall,', 'In conclusion,', 'In short,': a paragraph that announces it is summing up, then restates what the reader has just read. |
+| `false-range` | 1 | copy | warn | False range | 'From first-time buyers to seasoned investors alike' names two ends of no real scale, to sound as if it covers everyone. It describes nobody. |
+| `phrase-density` | 2 | copy | warn | Leaned-on phrase | 'Rather than', 'actually', 'That is...' openers, participle triads: each is ordinary English once, and a document that uses one far above its rate has one move and repeats it. |
+| `aphorism-density` | 2 | copy | warn | Aphorism cadence | Two clipped sentences alone on a line ('All of this already exists. The job is making it findable.') land once. Closing every section, they are a cadence a reader learns to hear. |
+| `contraction-scarcity` | 1 | copy | warn | No contractions | A long document with almost no contractions reads stiff and machine-made. People write 'we're' and 'don't'. |
+| `sentence-rhythm` | 2 | copy | warn | Metronomic rhythm | Sentence after sentence of the same length reads generated. A person's sentences run long, then short, then long again. |
+| `repeated-sentence` | 1 | copy | warn | Repeated sentence | The same sentence on two pages, or the same nine words lightly edited, is one of the strongest signs a document was assembled rather than written. |
+| `heading-shape` | 2 | copy | warn | One heading shape | Headings that all share one shape ('X, and Y', 'X, because Y', 'Keep, rewrite, consolidate, retire') read as a template filled in section by section. |
+| `heading-echo` | 2 | copy | warn | Heading echoed | A heading repeated word for word as the first line under it spends the reader's attention twice on the same words. |
 <!-- craft:catalogue:end -->

package/COPY.md

@@ -0,0 +1,280 @@
+# The house copy standard
+
+**Canonical.** Every client repo, every session, every piece of outbound
+writing. This file ships in `@domandigital/craft`, next to the tells that
+enforce it. Anything else claiming to be the copy rules is a copy: link here.
+
+- The rules are this file.
+- The detectors are craft's copy tells: `craft tells list` shows every one,
+  with why it is a default and what to do instead.
+- The house policy, which tells fail a commit, is `src/character/house.ts`.
+  `craft copy --gate` applies it. `copy-check` in claude-kit is a thin wrapper
+  over that, and the pre-commit hook runs it.
+
+## Why
+
+A reader who has seen a thousand generated pages recognises one in a line.
+None of the patterns below is AI-only, and people use all of them. The signal
+is **stacking**: one is plausible, three on a page reads as machine output.
+
+What makes copy read as written by a person is not the absence of tells. It is
+the presence of facts: a place, a time, a number, a name, who does the work.
+The rules below remove the defaults; the writing directives say what to put in
+their place.
+
+## Blocking tier
+
+A single hit fails the check. These are not judgment calls.
+
+- **Em dashes.** Use a full stop, a comma, a colon or a middot. En dashes for
+  genuine ranges (`Monday–Saturday`, `9am–5pm`) are correct and stay.
+- **Emoji** in shipped copy. Typographic marks are fine: `★` for a rating, `·`
+  as a separator, `→` in a link.
+- **Contrastive negation as a template**, in every form:
+  - copular: "It's not a website, it's a growth engine"
+  - verbal: "I don't teach the treatment you'd choose: I teach the standard"
+  - "not just X" / "not merely" / "not simply" / "more than just"
+  - "X isn't a job, it's art"
+  - "less about X, more about Y"
+  - parallel participle: "shipped from here, not imported from Seoul". Both
+    sides must be real participles, so "cosmetics, not medicines" (a plain
+    factual contrast) passes.
+- **A "No X" claim as the whole string.** "NO JARGON GUIDE", "No catch." A
+  short standalone "No X" reads as a slapped-on badge. Empty-state and error
+  text is exempt ("No results found", "No bookings yet"), as is anything under
+  a signed-in route (`/admin/`, `/portal/`, `/dashboard/`).
+- **"No X, no Y" as a list**, anywhere in a sentence. "No obligation, no
+  spam." Defining the business by what it is not, itemised.
+- The phrase list: `let's dive in`, `let's unpack`, `let's break this down`,
+  `here's the thing`, `it's important to note`, `at its core`,
+  `the key takeaway`, `the reality is`, `the truth is`,
+  `in today's rapidly evolving…`, `increasingly digital`,
+  `whether you're a X or a Y`, `seamless(ly)`, `unlock your…`,
+  `meaningful impact`, `foster collaboration`, `empower`, `leverage`,
+  `harness`, `delve`, `a comprehensive guide to`, `plain English`.
+
+Also blocking: `in the event that`, `at no additional cost to you`, `rest
+assured`, `it is worth noting that`, `when it comes to`, `we've got you
+covered`, `dive into`, `unlock the power`, `look no further`.
+
+**Buzzwords**: synergy, cutting-edge, best-in-class, best of breed,
+world-class, state of the art, game-changer, paradigm shift, mission-critical,
+low-hanging fruit, move the needle, circle back, touch base, reach out (use
+"get in touch"), value-add, robust, turnkey, holistic, disruptive.
+
+**Negative reassurance**: `no risk`, `never locked in`, `nothing goes wrong`,
+`no hidden fees`, `never pay`, `without the hassle`, `no surprises`. Do not
+name a fear and deny it. Say what they get: "The price on the quote is the
+price you pay", not "no hidden fees".
+
+`empower`, `leverage` and `harness` block because a plainer word almost always
+exists. Name the mechanism: "Your team can publish without us", not "we
+empower your team".
+
+`plain English` has one house replacement: **"properly explained"**.
+"Explained plainly" swaps a synonym and keeps the tell.
+
+## Review tier
+
+Printed, never blocks. One is normal English; several stacked is not. Only a
+person reading decides which instance earns its place.
+
+| Tell | What it catches |
+| --- | --- |
+| `review-phrase` | truly, genuinely, deeply, ultimately, great question, absolutely, one-size-fits-all, and superlative over-claiming: "the single most", "by far the most", "hands down the", "arguably the best", "bar none" |
+| `ai-vocabulary` | tapestry, testament, realm, elevate, nestled, meticulous, unparalleled, moreover; and underscore, showcasing, pivotal, measured 10 to 14 times over-used in post-2022 text |
+| `stock-phrase` | to the next level, one-stop shop, we pride ourselves, passionate about, peace of mind, "so here's", a journey used as a metaphor |
+| `hollow-imperative` | Discover, Experience, Transform, Elevate as the opening verb of a heading or button |
+| `rhetorical-opener` | "Looking for...?", "Tired of...?", "Ready to...?" |
+| `staccato-triplet` | "Fast. Friendly. Local." |
+| `where-x-meets-y` | "Where luxury meets comfort" |
+| `ing-tail` | "..., ensuring peace of mind", "..., highlighting our commitment": significance claimed, no mechanism |
+| `vague-attribution` | "studies show", "experts agree", "it is widely known": a claim credited to nobody |
+| `closing-summary` | "Overall,", "In conclusion,", "In short," opening a paragraph that restates what came before |
+| `false-range` | "from first-time buyers to seasoned investors alike", "everyone from students to retirees" |
+| `vague-word` | innovative, scalable, end-to-end, streamline, solutions. Reported only when a file is named outright |
+| `question-reveal` | "The result? Twice the bookings.", "What does this mean for you? It means...": a staged reveal |
+| `inline-label-list` | three or more bullets in a row opening "**Label:**", where the label repeats the line |
+| `placeholder` | "[Insert client name]", lorem ipsum, `TODO_PLACEHOLDER`. Fine in a draft commit; **zero in anything delivered** |
+| `chatbot-residue` | "Certainly! Here's a revised version", "as an AI language model", "let me know if you'd like me to", chat citation tokens, links tracked `utm_source=chatgpt.com`. **Zero in anything delivered**, and the first candidate for the blocking tier |
+
+### Style is not evidence
+
+Most tells here are house style: they make copy generic, whoever wrote it. The
+research is clear that an em dash, a rule of three or a rhetorical question
+proves nothing about authorship, and em dash rates even run in opposite
+directions between models. Only `chatbot-residue` is evidence that a chat reply
+was pasted. Report findings as an editorial diagnosis ("three benefit claims
+lack a mechanism; two stock phrases recur"), never as a score or an
+accusation.
+
+Two more review patterns no regex can separate from honest English, so they
+are for the read-through only:
+
+- **Trailing contrastive negation**: "Real Square data, not a vanity metric."
+  The same shape is how a genuine disclaimer is written ("general information,
+  not a recommendation").
+- **Abstract noun triads** with no concrete referent: "clarity, consistency,
+  confidence".
+
+## Density tier
+
+Review only, never blocks. Every rule above fires on one line read in
+isolation. These fire on a **rate**: each instance is legitimate English, and
+only the count gives it away. They read documents (Markdown, MDX, text, HTML)
+of 400 words or more, and the allowed count is a floor plus a rate per 1,000
+words, so a proposal is not held to a landing page's limit.
+
+Why they exist: a 30-page client proposal passed every per-line rule clean.
+A person then found "rather than" 54 times in 10,500 words, no contractions,
+one sentence pasted onto two pages, a heading repeated as its own first line,
+and an italic one-liner closing almost every page.
+
+| Tell | Fires above |
+| --- | --- |
+| `phrase-density` | "rather than" (4, +0.9 per 1,000 words), "instead of" (4, +0.7), "actually" (2, +0.35), "properly" (3, +0.4), "genuinely" (1, +0.2), "the single ...est" (1, +0.15), "That is / This is" openers (4, +0.7), "Every" openers (4, +0.6), "Worth noting" openers (3, +0.5), negation-led openers "None of / Not one / Nothing" (5, +1), participle triads "researched, written and referenced" (3, +0.6) |
+| `aphorism-density` | two clipped sentences alone on a line, "It already exists. The job is making it findable." (5, +1.2) |
+| `contraction-scarcity` | under one contraction per 1,000 words, in 800+ words. Formal legal pages are the deliberate exception |
+| `sentence-rhythm` | sentence lengths that barely vary: spread (standard deviation over mean) under 0.42 across 40+ sentences |
+| `repeated-sentence` | a sentence of nine words or more appearing twice, or two sentences sharing a nine-word run |
+| `heading-shape` | headings sharing one shape: "X, and Y", "X, because Y", a bare comma list (3, +0.5) |
+| `heading-echo` | a heading of four words or more repeated as the first words under it |
+
+The fix is never to delete every instance. Keep the ones that land hardest and
+vary the rest. That is why a density finding prints a count and an example,
+not a line to go and edit.
+
+## Writing directives
+
+These are what to do, not what to avoid. Read them **before** drafting. Do not
+read the tell lists before drafting: naming a phrase to avoid puts it in the
+writer's head, and it comes out anyway. Check against the lists afterwards.
+
+- **Lead with the fact.** A place, a time, a number, a name. "Gas Safe
+  registered since 2009. Same-day callouts in Brackley." beats "Trusted local
+  experts".
+- **Concrete nouns beat abstractions.** "Rated 5.0 from 58 reviews" beats
+  "trusted by the community".
+- **Name the mechanism.** Not "hassle-free booking", but "Book in three taps.
+  We confirm by text within the hour."
+- **Persuade with something checkable.** Sales copy has to sell, and
+  enthusiasm is allowed. Persuasion backed by a fact passes; persuasion with no
+  fact behind it is the tell.
+- **Concise, scannable, objective.** In usability testing, cutting word count,
+  making text scannable and dropping promotional tone each helped, and all
+  three together roughly doubled usability (Nielsen Norman Group).
+- **Vary sentence length.** Some long, some short, some flat. Uniform rhythm
+  reads generated, and so does uniform wit.
+- **Use contractions** where you would in speech. Formal legal pages are the
+  exception.
+- **If the sentence would read the same with a competitor's name in it**, cut
+  it or make it specific.
+- **Never reuse a sentence across two clients.** Templated language across
+  clients is a stronger tell than any phrase.
+- **Never open with a thesis-statement negation.** "This is a working build,
+  not a mockup." Judge by function: if the sentence exists only for the
+  contrast, cut it.
+- **Compliance hedging must not eat the sell.** Put trial design and caveats
+  where substantiation belongs (an evidence section, an ingredients panel, a
+  footnote). In the selling paragraph, name the claim and stop.
+- **End each major section with one action sentence**: what to do next.
+  "Book a call to see if we're a fit."
+
+## Register: British and professional
+
+The tool cannot grade register; a person does. British spelling and phrasing
+(colour, organise, licence as a noun, centre, towards, enquiries). Professional
+and confident, never matey: no over-familiar asides ("with me nowhere near
+it"), no American idiom ("shallow wins", "looks busy"), no "honestly" as an
+intensifier. Do not change a brand name, domain or legal entity wording.
+
+## Word limits
+
+| Element | Limit |
+| --- | --- |
+| H1 | 10 words |
+| Subhead | 24 words |
+| Paragraph | 2 sentences |
+| Bullet | 7 words |
+
+Hyphenated compounds count as one word. Not automated.
+
+## Proof rule
+
+- **No invented results, testimonials, logos, certifications or numbers.
+  Ever.**
+- Every proof point comes from the project's own proof source (a proofs or
+  reviews file, the CMS, the client). If one is missing, write
+  `TODO_PLACEHOLDER: <the exact field needed>` and say so. Never fill it.
+- A quotation under a named person must be theirs. Copy drafted for someone,
+  in quote marks with their job title beneath it, is a fabricated attribution.
+
+## CTA rule
+
+- One primary CTA label site-wide on commercial pages, in the form **action +
+  outcome**: "Book a free strategy call".
+- The secondary CTA is the same everywhere it appears.
+- Labels and links come from the project's config, never retyped per page.
+
+## Never edit to pass: quoted words
+
+**A real person's own words are exempt from every rule here.** Reviews,
+testimonials, a practitioner's bio, a quote from the client. Editing them to
+pass an audit falsifies a quotation, which is far worse than an em dash. Mark
+the line `copy-ok` or add the file to `.copyauditignore`, and leave it as they
+said it.
+
+**Owner-editable copy is advised on, never overridden.** Where a client types
+their own words into a CMS, report what the check finds and tell them. Copy
+that ships from the repo is ours to fix; copy the owner typed is theirs.
+
+## Rendered social assets: no raw links
+
+A carousel slide, Story, Reel or caption never shows a raw URL: no `https://`,
+no `www.`, no path. Show a bare domain (`RMP-ELECTRICAL.CO.UK`) or a plain
+call to action ("Link in bio"). Keep the full address in source data and let
+the render step transform it. craft cannot enforce this, because raw URLs are
+legitimate almost everywhere else in a repo. The social generator's own render
+gate enforces it, before its first render ships.
+
+## Suppressing a finding
+
+For genuine exceptions only, and the reason matters more than the mechanism:
+
+- `copy-ok` (or `craft-ok`) on the line or the line above: a one-off. Listed
+  in every report, never hidden.
+- `.copyauditignore` at the repo root: globs, matched on the repo-relative
+  path and the bare filename.
+- `.nocopycheck` at the repo root: skips the repo.
+- An `exceptions` entry in `art-direction.json`, with a `because` of at least
+  a sentence.
+
+Reaching for these more than occasionally means a rule is miscalibrated. Fix
+it here and in craft, not in forty repos.
+
+## Rewriting without losing facts
+
+A rewrite that removes every tell and drops the price is worse than the
+draft. Before handing over any rewrite, compare it with the original:
+
+```bash
+craft copy compare original.md rewrite.md
+```
+
+It lists every protected fact (prices, numbers, dates, times, phone numbers,
+emails, links, postcodes, names) the rewrite **lost**, and every one it
+**added**. Exit 1 if either list is not empty. Restore a lost fact or say why
+it went; source an added fact or remove it, because an unsourced number is an
+invented one.
+
+## Changing a rule
+
+1. Edit this file.
+2. Add or change the tell in `src/character/tells/`, with a flag fixture per
+   detection path and a pass fixture that is real persuasive copy. A tell that
+   fires on good sales copy does not ship.
+3. Place it in `src/character/house.ts`. New tells start as `review`. A tell
+   moves to `block` only after its hits across the estate have been read by a
+   person and the baseline is recorded in `ROADMAP.md`.
+4. `pnpm --filter @domandigital/craft test`. The binding test fails if a
+   phrase in this file's blocking tier does not block.

package/README.md

@@ -94,6 +94,18 @@ reason, in `art-direction.json`:
 { "exceptions": [{ "tell": "ai-violet", "because": "Violet is on the van, the cards and the fascia." }] }
 ```
 
+### Copy
+
+[COPY.md](./COPY.md) is the house copy standard: the blocking tier, the review
+tier, the density tier that reads whole documents, and what to write instead.
+`craft copy --gate` applies the house policy, so the blocking tier fails the
+run. A test fails if a phrase the standard blocks does not block.
+
+```bash
+npx craft copy --gate content   # the house gate: exit 1 on the blocking tier
+npx craft copy compare draft.md rewrite.md   # facts a rewrite lost or added
+```
+
 The same checks run in code:
 
 ```ts

package/dist/{audit-YNODYB3U.js → audit-W57YDZ2K.js}

@@ -6,9 +6,9 @@ import {
   snapshotPage,
   snapshotUrl,
   snapshotUrls
-} from "./chunk-OA7SGPN4.js";
-import "./chunk-Q54JSNC5.js";
-import "./chunk-3AP4PJMX.js";
+} from "./chunk-Q4MN3OXF.js";
+import "./chunk-OGSYVCMU.js";
+import "./chunk-CZHRNPDD.js";
 export {
   audit,
   collectInPage,