v2 differs from legacy/new only by the callback dialog; the calculator and
the new phone number were already in the React landing.
The header «Звонок» button opens the «Перезвоним вам» dialog from the
fitness landing (name, phone, consent). Unlike fitness, the phone field
keeps this landing's +7 mask and refuses fewer than 11 digits, as the
mockup does. The server accepts form 'callback', names the deal
«Обратный звонок — <имя>» and tags it «обратный звонок». The phone link
inside the dialog and in the section sheet still reports a call click.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Ports legacy/hotel v7 into the React landing.
- Copy: nav is Экономика / Для гостя / Интерьер / Сервис, and «Интерьер»
now comes before «Сервис». The hero keeps one button and gets the
2 500 ₽ / 600 000 ₽ / 7,2 млн ₽ facts. New headings and leads for the
interior, service and request blocks. Session lengths 10–15 and 30–60
min, area 12–36 m².
- Calculator counts sessions and gains the ЭкзоКлиник benchmark and a
«Получить консультацию» button.
- Request form: company and email are optional. The server schema
already treats them as optional.
- v8 phone type scale (≤620px), a 380px `micro` step, and a `slim`
(≤640px) variant. The max-width variants are now declared widest first,
so the narrower one wins where two apply.
- The market charts' SVGs stay inline, as in the mockup.
The header «Звонок» button opens the «Перезвоним вам» dialog from the
fitness landing (name, phone, consent). The server accepts form
'callback', names the deal «Обратный звонок — <имя>» and tags it
«обратный звонок». The phone link inside the dialog still reports a call
click.
With Inter substituted into the mockup, the page text matches it line for
line (except the fixed «2025–2026 гг.» typo), and every section height
matches at 360–1440 px.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
legacy/Fitnes v2/ is the static mockup the landing was just ported from
(README points to it as the current visual reference). launch.json gains a
fitnes-v2-legacy entry that serves it on :8090 for side-by-side review.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Ports legacy/Fitnes v2 into the React landing.
- Copy: nav is now Экономика / Оборудование / Программы / Интерьер, every
CTA reads «Получить консультацию», the hero has one button, a new label
and new facts.
- Calculator: the average check defaults to 2 500 ₽, and a new investment
field drives payback and 12-month ROI. Adds the ЭкзоКлиник source note
and a new disclaimer.
- Constructor: «Состав оборудования», indications on the device cards. The
CTA stays live with fewer than three devices picked, and the copy points
to a consultation instead.
- «Юридическая рамка» is replaced by «Форматы размещения» (12 / 24 / 36 м²).
- Lead form: city and club format are dropped; company and email are
optional. The server schema already treats them as optional.
- Section headings share one compact scale (headingScale + a `slim`
640px variant), as in the design's v7 rule.
Deliberately kept from the current site rather than the mockup: the
+7 927 789-60-71 number and the callback modal behind the header call
button, both newer than the design.
With Inter substituted into the mockup (it ships no font and falls back to
Arial), every section height matches at 900 and 1440 px. At 375 px the
hero is 6 px shorter because of the existing fluid h1 on phones.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The counter was in place but nothing reached it: src/lib/lead.ts only pushed
a custom dataLayer event, which Metrika's ecommerce: 'dataLayer' ignores, so
form conversions were not counted at all.
All five forms — including the fitnes callback modal — funnel through the one
success branch in submitLead(), so a single ym(112352796, 'reachGoal',
'FORM_SUCCESS') per landing covers every one of them. The counter id is
hardcoded to match the inline snippet in index.html; two sources for it would
drift. The call is optional-chained and wrapped in try/catch, because losing
analytics must never cost a lead.
The honeypot is the one success that is not a conversion: the server answers
bots with 200 {ok:true, leadId:0} so they learn nothing, and leadId === 0 is
what keeps that answer out of the goal.
The goal itself still has to be created in the counter's settings as a
JavaScript event named FORM_SUCCESS — noted in deploy/README.md, replacing
the entry that said conversions were unwired.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The published number becomes +7 927 789-60-71 across all four landings —
`contacts` in each `src/data/content.ts`, plus the fallback messages in the
lead endpoints and `src/lib/lead.ts` that spell it out when amoCRM is down.
The `legacy/` reference pages keep the old number: they are the visual
originals, not something that ships.
Tapping the phone button now also creates a lead. It cannot reuse the form
path: a click carries no name and no number, so `callClickSchema` in
`shared/lead.ts` validates the tracking data alone, `LeadService.submitCallClick`
creates a contactless lead tagged `клик по телефону`, and `AmoClient.createLead`
takes an optional `contactId` for it. The mapper's field/note split moved into
`collect()` so both lead kinds share it.
The browser fires this with `sendBeacon` (falling back to `fetch keepalive`),
because the same click hands the page to `tel:` and a plain fetch would be
cut off mid-flight. Duplicates are held down from both ends: one lead per
browser session on the client, four per IP per 30 minutes on the server.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Counter 112352796 goes into the <head> of every landing's index.html, ahead
of the Vite entry, so the hit fires before React mounts. The pages are plain
SPAs served off disk by nginx (Express only answers /api/*), so the static
head is the only place it needs to be.
One counter serves all four domains, which means reports mix them and every
domain has to be listed in the counter's own settings or its hits get
filtered — noted in deploy/README.md along with the conversion tracking that
is still missing: src/lib/lead.ts pushes a plain custom dataLayer event,
which Metrika's ecommerce: 'dataLayer' does not read.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Found while verifying the first real provision: none of the four
security headers reached the browser on any HTML page.
nginx's add_header inheritance is all-or-nothing — a location that sets
any add_header of its own discards every header inherited from the
server block. "/" resolves through try_files to `location = /index.html`,
which sets Cache-Control, so HSTS, X-Content-Type-Options,
X-Frame-Options and Referrer-Policy were silently dropped exactly where
they matter. /assets/ lost them the same way.
Move the four into snippets/security-headers.conf and include it in the
server block and in both locations that add a header of their own.
Also from the same provision run:
- ssl_stapling is dead config now that Let's Encrypt certificates carry
no OCSP responder URL; it only logs a warning per cert on each reload;
- README step 3 chmod'ed /etc/exo to 750 but never set its group, so the
exo user could not traverse it and exo-deploy died on "cannot read
/etc/exo/<app>.env";
- README step 7 dropped http-extras.conf into conf.d without disabling
the same directives in Debian's stock nginx.conf, and nginx refuses to
start on a duplicate gzip / server_tokens.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Bring deploy/ in line with the renamed app directories and with the
domains that are actually configured in the per-app .env files:
fitnes fitness.exorecovery.ru 3000 fitness-landing
hotel hotel.exorecovery.ru 3001 hotel-landing
medcenterphysio physio.exorecovery.ru 3002 physio-landing
medcenterstart start.exorecovery.ru 3003 start-landing
apps.conf, the README provisioning steps (directories, bare clones,
systemd units, the sudoers whitelist) and the exo-deploy usage line all
used the old medcenter / medcenterpersonal keys. The README table and
app.build.env.example still carried the *_DOMAIN placeholders.
Also refresh the tracked .env.example files: real VITE_SITE_URL values
instead of the old exodevices.ru sub-paths, and the amo lead tags that
match apps.conf.
Two lockfile fixes, both of which broke `npm ci` in exo-deploy:
- medcenterstart had no lockfile at all after package-lock.json and
pnpm-lock.yaml were removed — restore package-lock.json and keep the
pnpm one deleted, since the deploy path is npm;
- medcenterphysio/package-lock.json had lost "resolved" and "integrity"
on 147 of its 236 entries — restore the complete file.
Verified: exo-render-nginx renders all four vhosts with no leftover
placeholders, and every package-lock.json agrees with its package.json.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Rename the two medcenter landings to their audience names:
medcenter -> medcenterphysio, medcenterpersonal -> medcenterstart.
Alongside the rename:
- add a RevenueCalculator section to both landings;
- rework the copy and figures in src/data/content.ts;
- simplify the lead form: drop the "cabinet_state" and "profile"
selects (along with SelectField and the matching fields in
shared/lead.ts, lead-mapper.ts and amo-check.ts) and make
company and email optional;
- add the legacy/new static prototypes for both landings;
- add pnpm-lock.yaml to medcenterstart (package-lock.json is still
there too).
deploy/apps.conf and deploy/README.md still refer to the old
directory names and need a follow-up.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The repo had no infrastructure at all: no Dockerfile, no nginx config, no
systemd unit, no deploy script, no CI. Deployment guidance existed only
as two prose lines in each README.
Target shape is one reg.ru KVM box serving all four landings, each on its
own domain root. nginx serves the built SPA off disk and proxies only
/api/* to a small Express process per app, so a crashed or restarting API
takes down the form rather than the page. Releases are built on the
server into a timestamped directory and swapped in by renaming the
`current` symlink.
deploy/apps.conf domains and ports, the one place to edit
deploy/README.md hardware/software spec, provisioning, runbook
deploy/bin/ exo-deploy, exo-render-nginx
deploy/nginx/ vhost template, shared snippets, http extras
deploy/systemd/ one template unit for all four instances
deploy/env/ annotated templates for the two env files
exo-deploy asserts the VITE_SITE_URL substitution actually happened,
prunes devDependencies after building, precompresses for gzip_static,
health-checks after the swap and rolls back on its own if the new release
fails to answer. exo-render-nginx refuses to emit a vhost while a
placeholder domain is still in apps.conf.
Secrets are split across two env files: the amoCRM token stays in
/etc/exo at mode 640 and is injected by systemd, while only public
build-time vars are copied into a release. dotenv does not override
variables already in the environment, so the two coexist safely.
Chose systemd over Docker deliberately — no database, no conflicting
runtimes, no CI or registry to build images in, and containerising fights
the serve-static-from-disk design. Reasoning and the triggers to revisit
it are recorded in the README.
Domains are still placeholders; nothing here has been run against a
server yet.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Fixes that block or weaken a real deployment. Nothing here changes the
rendered pages.
Bind the lead API to loopback. Each server called app.listen() without a
host, so it bound 0.0.0.0. Combined with a blanket `trust proxy: true` —
which makes Express take the leftmost X-Forwarded-For entry as req.ip —
the in-memory lead rate limiter was spoofable by anyone who could reach
the port directly. HOST now defaults to 127.0.0.1 and trust is narrowed
to 'loopback', so a request arriving from anywhere but the local proxy
has its forged header ignored.
Give each landing its own port. All four .env files claimed PORT=3000,
and fitnes/.env.example collided with medcenter/.env.example, so three of
the four could never have started on one host. Now 3000/3001/3002/3003
consistently across the code defaults, the env templates and the vite
dev proxies, so all four also run side by side locally.
Template the JSON-LD url. canonical and og:url already resolved from
%VITE_SITE_URL%, but the JSON-LD block hardcoded an exodevices.ru
sub-path that would not follow the environment. All four now read from
the same variable.
Declare the Node version. Nothing stated it, yet transitive deps impose
a >=22.12 floor (@rolldown/binding, yargs, concurrently). Added engines
and .nvmrc so a too-old runtime fails clearly.
Typechecked and production-built on all four; verified the socket binds
127.0.0.1 only, health reports amo:true, the site still boots with the
CRM unconfigured, and the limiter returns 429 with Retry-After on the
ninth request.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Four defects, all of them real clipping rather than styling taste.
The header's two action pills were `shrink-0` at 288px inside a 343px
container, so flexbox took the difference out of the logo, which carried
`min-w-0`. The brand mark rendered 37px wide at 375px and 0px at 320px,
and the CTA ran past the right edge. The wordmark stays legible now and
the call button goes square and icon-only below 760px; the CTA moves to
a new MobileActionBar, which has room for it and puts it in the thumb
zone. It waits for the hero's own buttons to scroll away and drops back
down while the form is on screen, so it never covers the fields it
points at.
`.scroll-cards` bleeds to the viewport edge with a 16px inline padding
and a matching negative margin, but had no `scroll-padding-inline`, so
`scroll-snap-align: start` resolved to scroll position 16 and mandatory
snapping forced that scroll on load: the first card sat on the bezel
with no gutter, and one row had drifted a full card over. Matching the
padding fixes the resting position on all seven rows.
Card tracks move from a percentage of the row to `calc(100vw - 62px)`.
A percentage peek grows with the screen, and at 82-90% it exposed ~68px
of the next card -- enough legible half-words to read as broken text.
The value stays in the markup as `auto-cols-*` so it keeps its sort
position against `sm:grid-still`, which has to win at the breakpoint.
The market stat card pinned its source line to the card bottom while the
longest label wraps to four lines at phone width, running text through
text by 75px. The source is in flow under `mt-auto` instead.
The header CTA switches on `max-md` rather than the `narrow` variant:
`narrow` is max-width 760 and the bar is min-width 760, so at exactly
760px both would have hidden and the page would have had no CTA at all.
Verified at 320 / 375 / 759 / 760 / 800 and desktop: no horizontal page
scroll, no in-flow overflow, desktop grids unchanged.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Three layout bugs and a mobile experience that had no wayfinding.
The `#proposal` section was clipped, not merely cramped: the submit
button carried `whitespace-nowrap` from the shared button base, giving
it a min-content width of 358px, which sized the section's implicit
`auto` grid track to 406px inside a 343px container. `overflow-hidden`
then sliced the copy and every form field off the right edge of a 375px
screen. Fixed at the root — `whitespace-nowrap` moves out of `base` and
into `headerSize`, where buttons are sized by their label — plus an
explicit `grid-cols-[minmax(0,1fr)]` on the section.
The market stat cards put their source line in absolute position against
a guessed min-height, so the longest label ran underneath it. It is in
the flow now, pinned by `mt-auto`.
Below 980px the header nav is not rendered, which left seventeen screens
of scroll with no way to navigate them. The header gives that width back
to the page — logo and call button only — and <ThumbBar /> takes over:
scroll progress, a live section indicator that opens a section sheet,
and one short CTA, hidden over the hero and over the form.
Card rows now say what they are. Rows of unordered peers keep scrolling
sideways and gain a measured segment indicator (<Rail />), replacing the
two prose "swipe sideways" hints. The launch timeline goes the other
way: sideways scrolling hid week 3-5 from someone reading week 1-2, so
below 900px it becomes four disclosures on a spine — which also recovers
the scroll length the taller hero spent.
Below 560px the type scale steps down half a level and the display
tracking opens back up, and the hero photo zooms past `cover` onto the
wave wall, since fitting by height left only ceiling tiles above the
copy.
Also: the route card inside the market section split into two 165px
columns at 760px, so that split now waits for 980px.
Desktop above 980px is unchanged.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Ports the last single-file page, MedCbezFiz — the medical centre that has no
physiotherapy yet — to the architecture the other three landings now share:
Vite + React 19 + TypeScript + Tailwind v4, Express 5 for the API, zod schema
shared between client and server.
The original is kept in legacy/index.html as the visual reference, with its
instructions alongside it in legacy/README.md. Its 4 MB collapse to 317 KB of
JS plus assets loaded on demand. Fourteen inlined images were extracted; twelve
turned out byte-identical to the medcenter landing's, and the two large device
PNGs (1.5 MB and 750 KB) had identical sources, so the webp conversions from
that landing are reused rather than redone — 664 KB of assets instead of 2.8 MB.
Text was verified rather than eyeballed: the app is server-rendered to static
HTML and its visible text diffed against the legacy file. 280 blocks against
281, and the only difference is the honeypot's screen-reader label, which the
original has no equivalent for.
Two things in the original are dead and were deliberately not ported: 39
.patient-economics / .base-economics-card rules that no element uses, and the
[data-counter] animation handler, which likewise matches nothing on the page.
Both are documented in the README.
Departures from the medcenter landing this was built from, all following the
original: the hero photo is anchored at 65% (72% from 980px), the header is
transparent until 18px of scroll instead of always painted, the revenue band
keeps 32px gutters at every width, and the launch timeline breaks into two
columns at 900px rather than 760px.
Leads post to /api/leads/medical-centers-no-physio, the address the original
page already referenced, into pipeline 10980758. The cabinet_state select has
no counterpart on this form and was dropped from the schema, mapper and check
script. Vite runs on 5175 and the API on 3002 so all four landings can run at
once.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Ports the single-file MedCsFiz page — the medical centre that already runs
physiotherapy and wants the existing room to earn more — to the same
architecture as the fitness and hotel landings: Vite + React 19 + TypeScript +
Tailwind v4 on the client, Express 5 for the API, zod schema shared between the
two.
The original stays in legacy/index.html as the visual reference. Its 4 MB of
inlined CSS, JS and base64 images become 12 asset files (664 KB) plus a bundle
loaded on demand; the two 1.5 MB / 750 KB device PNGs are recompressed to webp.
The page is split into 18 components with all copy moved to src/data/content.ts.
Leads reuse the fitness amoCRM integration: contact lookup by phone in every
Russian spelling, deal in the first stage of pipeline 10980758, account fields
matched automatically with the rest written to a note. Rate limit, honeypot and
a localStorage fallback so a lead survives the CRM being down. Endpoint is
/api/leads/medical-centers-existing-physio; Vite runs on 5173 and the API on
3000.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Ports the single-file ЭкзоОтель page to the same architecture as the fitness
landing: Vite + React 19 + TypeScript + Tailwind v4 on the client, Express 5 for
the /api/leads/hotels endpoint, zod schema shared between the two.
The original page is kept in legacy/index.html as the visual reference. Its 15
inlined base64 images are extracted to files (the 1 MB HTML becomes ~318 KB of
JS plus assets loaded on demand), and its text is reproduced line for line —
verified with an innerText diff. Section heights stay within 0.6% at 375 and
1440 px; the drift comes from Inter actually loading, which the original asked
for but never served.
Three deliberate departures, documented in the README:
* eyebrow and lead in the CTA block were dark teal on navy (3.4:1); they now
match the other dark sections
* hero fact values overflowed their 80px column into the label below 430px
* "2025–2026г.." typo in the market source note
Leads reuse the fitness amoCRM integration: contact lookup by phone in every
spelling, deal in the first stage of pipeline 10980758, account fields matched
automatically with the rest written to a note. Rate limit, honeypot, and a
localStorage fallback so a lead survives the CRM being down. Vite runs on 5174
and the API on 3001 so both landings can run at once.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Port the single-file landing (2.6 MB of inlined CSS, JS and base64 images,
kept as fitnes/legacy/index.html) to Vite + React 19 + TypeScript, with an
Express API that files every form submission into amoCRM pipeline 10980758.
- Extract the 14 embedded images to src/assets/images and public/
- Rebuild the design system as Tailwind v4 @theme tokens; the stock palette and
breakpoints are cleared so only the EXO scale is reachable from utilities
- Split the page into 15 components; all copy moves to src/data
- Lead endpoint: find-or-create the contact (Russian phone spellings compared on
the last 10 digits), create the lead in the pipeline's first stage, map the
fields the account already has and put the rest in a note. If amoCRM is
unreachable the payload is logged and kept in localStorage rather than lost.
- Add a callback modal as a second entry point, tagged separately in the pipeline
- Self-host Inter Variable so the layout's 760/850/900 weights render as real
weights instead of snapping to bold
Fidelity was checked by comparing section offsets and heights against the
original at 375/480/640/900/1120/1440 px; every section and the total page
height matched exactly. Loading Inter deliberately changes text metrics, so the
byte-exact comparison holds against the pre-font build.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Yuriy PanovandClaude Opus 5
yuriy.p