GMedAC is not an emergency service and cannot respond to one. There is no queue, no on-call rota, no monitored channel and nobody watching a page load at three in the morning. The marketplace is at Phase 0 — no checkout, no live payment provider and no deployment — and this journal is a static site with no database and no way to receive a message. A reader in difficulty who treats any page here as a way to reach help has already been failed by that page. Nothing in this publication is a route to help.
The publication must never read as triage
Triage is a judgement about urgency made by somebody who can see the patient. This publication
sees nobody. output: export makes the framework refuse cookies(), headers(), searchParams
and middleware at build time, and search runs in the reader's browser against a static JSON
index, so nothing about the person reading reaches anything that renders. A page that sorted
readers by urgency would be sorting on nothing.
The following are refused by name. No check reads an article body for them, so the refusal is an editorial one: naming them is what makes a later editor argue against a list rather than into a silence.
- Checklists that resolve to a verdict on whether a situation is an emergency.
- Severity scales the reader scores about themselves, and decision trees that arrive at the same verdict by a longer route.
- Any wording whose effect is to discourage contact with emergency services, including reassurance offered in the absence of an assessment.
- Any framing that offers the article as a substitute for contacting emergency services.
The publication can describe how a service is organised in a place. It cannot tell a reader whether they need it.
Emergency information is jurisdiction-specific
Emergency numbers, service names, who answers, what is dispatched, what it costs and who is
eligible differ by country and often below country level. The United States is never one
corridor; it resolves state by state, and the US material in the marketplace record is corridor
discovery — a shortlist and a source register — not a clearance. No launch corridor is approved
anywhere, and EVIDENCE_MISSING is not a clearance.
An article carrying such material declares emergencyInformation: true, names in prose the
jurisdiction it describes, and carries validityAsOf and nextReviewAt like every other
article. A jurisdiction-specific claim resting on REASONED_INFERENCE,
COUNSEL_CONFIRMATION_REQUIRED or UNKNOWN cannot publish — the claims check refuses it, and
that refusal is worth more here than anywhere else in the repository. The absence of a
jurisdiction from this journal says only that nobody has researched it. Excluded means
unresearched. It never means permitted.
Official resources, cited like everything else
Emergency material rests on LEGISLATION, REGULATOR_PUBLICATION, OFFICIAL_REGISTER or
CLINICAL_GUIDELINE sources, recorded in sourceIds with dateRetrieved and a prose
retrievalRoute. That restriction is a requirement on the editor: nothing keys a source check to
emergencyInformation. What is enforced is narrower — an article declaring HIGH or CRITICAL risk
may cite nothing weaker than PEER_REVIEWED_STUDY, which excludes INSTITUTIONAL_SUMMARY and
JOURNALISM — so a number reproduced from a news report is stopped by the build only where the
declared risk reaches that floor. retrievalRoute is prose because a successful HTTP status is
not proof that anybody read the source, and a link that resolved is not a number that answers.
Interactive symptom triage is not built and will not be
There is no symptom checker, no urgency calculator and no interactive assessment of any kind, and
none is planned. Content is Markdown rendered through remark/rehype with rehype-sanitize rather
than MDX, so an article cannot execute JavaScript and no such thing can arrive as content. That
stops a checker written into an article. It does not stop one built into the site as a component,
and nothing but this policy stops the second. So the refusal is cheap to hold, which is not why it
is held.
It is held because a checker that is wrong for one reader in a thousand is wrong in exactly the
case the reader opened it for, and because building one would make GMedAC an assessor of
patients. The professional retains full clinical independence over whether to accept a patient
and over everything clinical that follows. GMedAC holds no clinical record and issues no clinical
document, and VERIFIED is a statement that somebody checked a document, never permission to
practise or to book.
Staleness fails the build at any risk level
A stale article declaring emergencyInformation: true fails the production build whatever its
contentRisk. This is the strictest freshness rule in the repository, and it is deliberately
harsher than the rest of the freshness policy, where LOW and MODERATE staleness is a warning and
an editorial-queue item rather than a failure — failing the build on an explainer a week past its
date teaches people to push the date rather than read the article.
That reasoning does not survive contact with emergency material. The reader who needs it is the least able to notice it is wrong. A reader consulting an explainer over coffee can cross-check a detail that looks odd; a reader in a crisis takes the first number on the page and uses it. A wrong number is not a degraded page, it is a delay measured in the only units that matter here. The cost of a stale emergency page is not paid by the person who let it go stale.
Emergency content is not created in order to have articles
A crisis page nobody is maintaining is worse than no page, because it looks like maintenance. So emergency articles are not written to fill a section, to give a new locale a plausible table of contents, to look complete at launch, or because the search demand is obvious and cheap to serve.
An emergency article is commissioned only where a named editor and a named reviewer have committed
in advance to its review cadence. No check can see that commitment; the build sees only the date it
produces. Where a review is missed and cannot be done, a person moves the article to WITHDRAWN
rather than leaving it to wait for attention — nothing moves it by itself, and what the build does
instead is refuse to ship it stale. Withdrawal is complete rather than cosmetic, because only
PUBLISHED and UPDATED reach a page, a sitemap, a feed or the search index. A material
correction to such an article is refused without a reviewer, and the correction file keeps
previousClaim verbatim beside correctedClaim. An empty emergency section is an acceptable
state.
What this policy does not catch
emergencyInformation is declared by the editor, exactly like contentRisk. An article that
prints an emergency number but declares the flag as false receives ordinary freshness handling,
and the strictest rule in the repository never runs on it. The build cannot read prose for a
telephone number and does not pretend to. Nothing downstream repairs this, because everything
downstream keys off the same declared field. What stands against it is a second person in
EDITORIAL_REVIEW, named authors and reviewers on the page, and the fact that this paragraph
exists. This is honest, not clean.