DSRPTV INLINE-PLATFORM PROBE v1.0 — DIAGNOSTIC PAGE. DO NOT PUBLISH. DELETE AFTER TESTING.

Platform: ____________________  ·  Rooftop: ____________________  ·  Franchise: ____________________  ·  Date: ____________

TEST A — Inline style survival and var() resolution

The box below reads three different results by color. This single test distinguishes all three states.

GREEN	Inline styles survive AND CSS custom properties resolve. Full token model is available. Best case.
RED	Inline styles survive but custom properties are sanitized. The var() fallback carried it. Build with hardcoded values per rooftop; fallbacks are load-bearing.
NO FILL	Inline style attributes are stripped. The token model does not apply, and a class-based variant is required for this platform.

Nested-inheritance check — this paragraph's parent carries the token, and the swatch below sits three levels deep:

TEST B — <style> block survival

A <style> block at the top of this paste targets the box below and nothing else.

RESULT BOX B — if this is GREEN with white text, <style> blocks survive. If it is plain/unstyled, they are stripped and the production ban holds.

A survived <style> block would unlock hover states, media queries, and :nth-child striping — all currently unavailable. Worth knowing even though our default posture stays inline-only.

TEST C — Is a CSS framework loaded in the content area?

These blocks carry NO inline styling. Anything that renders styled is coming from a framework the platform loads.

C1 — Bootstrap 3 only markers

label label-default — a small gray pill here means BOOTSTRAP 3.

C2 — Bootstrap 4 only markers

badge badge-pill — a rounded pill here means BOOTSTRAP 4.

C3 — Framework-agnostic grid

A / B / C side by side = a Bootstrap grid IS loaded in the content area. Stacked vertically = it is not, and all multi-column layout must be inline flex. Both C1 and C2 unstyled + C3 stacked = no framework at all, which is the expected DEP result and the assumption to verify on DealerOn.

TEST D — Inline flexbox, our grid substitute

Three across on desktop and stacked on a phone = display:flex and flex-wrap both survive. This is the only responsive mechanism available when <style> blocks are stripped, so check this one on a real phone, not just a narrowed browser window.

TEST E — Wide table and the overflow-x scroll wrapper

On a phone: the table scrolls sideways inside its own box and the page does not = wrapper survives. The whole page scrolls sideways = overflow-x:auto was stripped, and there is no mobile table strategy on this platform. Row banding present = inline row backgrounds survive.

E2 — Bootstrap table utility classes (no inline styling)

Header A	Header B
Striped?	value
Bordered?	value
Third row	value

Alternating row shading and grid lines here = table utility classes exist (framework loaded). Plain table = they do not, and banding must be inline row by row.

TEST F — <details> / <summary> survival

RESULT F — click to expand

If you can see this only after clicking, <details>/<summary> survives and native no-JavaScript accordions are available on this platform.

A clickable disclosure that expands = survives. Both lines visible at once with no control, or the summary text gone entirely = stripped or flattened. This is the only route to an accordion on a platform that strips JavaScript, so it matters for FAQ and long-form pages.

TEST G — Inline SVG

RESULT G

A green checkmark above = inline SVG renders and stroke="currentColor" inherits correctly. Nothing, a broken glyph, or a black square = SVG is stripped or mangled, and the styled-card treatment is the only icon substitute.

TEST H — Is an icon font loaded?

FA4 syntax:  |  FA5 syntax:

Recognizable glyphs after "FA4 syntax" = Font Awesome 4 is loaded. Glyphs after "FA5 syntax" = Font Awesome 5+. Empty space or hollow boxes in both = no icon font, which is the expected DEP result. Note WHICH set rendered — FA4 and FA5 syntax are not interchangeable.

TEST I — Does the theme force heading sizes?

The three headings below carry NO inline font-size. The gray reference bars are inline-styled to exact pixel sizes. Compare.

Bare H1 — no inline size

Reference: 32px

Bare H2 — no inline size

Reference: 26px

Bare H3 — no inline size

Reference: 19px

Watch for the Apollo defect: if the bare H1 renders SMALLER than the bare H2, the theme is forcing H1 down with !important and this platform needs the dual-element title pattern. Also note the body base size in DevTools — if it is a predictable 16px, the style guide's em type scale becomes usable instead of absolute px.

TEST J — id attribute and fragment anchors

Click here — it should jump to this section's heading.

Jumps correctly = id survives the save cycle and H2 fragment anchors work. Does nothing = id was stripped, and jump navigation is unavailable. Confirm in DevTools that id="probe-anchor-target" is actually present on the H2.

TEST K — What schema does the platform inject on its own?

This page contains ZERO structured data by design. Anything Rich Results Test finds is the platform's own, and is what our schema artifact must omit or reference by @id.

Entity	Emitted?
AutoDealer
LocalBusiness
AutoRepair
BreadcrumbList
WebSite
Organization

Also check in DevTools: is a <link rel="canonical"> auto-emitted? Does the page template render its own <h1> outside the content area? Both change what we ship.

Separately, paste a bare <script type="application/ld+json">{}</script> into the BODY editor on a second throwaway page and save. If it survives there, schema delivery is simpler than assumed. Expected result: stripped, and it belongs in the header / custom-code injection field.

TEST L — Image handling

?? Before pasting, replace the placeholder below with one real image URL from a CDN we actually use. No URL is hardcoded here on purpose — we never ship invented URLs, not even in a probe.

Renders at container width and scales down on a phone = external CDN images render and width:100%;height:auto;display:block is a working responsive substitute. Renders at native width or overflows = the inline sizing was stripped.

TEST M — Does the platform rewrite our links?

tel: link — authored by us  ·  external link with target and rel

In DevTools, confirm the tel: href came back as tel: and was NOT rewritten to wtai:// or wrapped in a tracking redirect, and that target / rel survived. Some platforms rewrite phone links at render time, which silently breaks click-to-call on mobile.

Step 3 — Source readback checklist

Reopen SOURCE view after saving and confirm each item. A visual pass cannot tell you whether an attribute was removed or merely overridden — only the returned markup can.

• Wrapper style attribute present, with the --probe-token declaration intact
• style attributes present on tables, table cells, and anchors
• <style> block present or removed
• class attributes preserved (they are useless without a framework, but note it)
• id="probe-anchor-target" still on the H2
• <details> / <summary> tags intact, not flattened to divs
• Inline <svg> intact, attributes not reordered or dropped
• <table> / <thead> / <tbody> structure unchanged
• Nothing re-wrapped in editor-added <p> or <div> tags
• Nothing reordered relative to the paste

END OF PROBE — record results, then DELETE this page.