diff --git a/bible-local/README.md b/bible-local/README.md
index 2b1b8bc..eca6cda 100644
--- a/bible-local/README.md
+++ b/bible-local/README.md
@@ -16,6 +16,7 @@ Project-specific architecture and operational contracts.
 | [09-vendor-spec.md](09-vendor-spec.md) | Vendor BOM and CFXML import contract |
 | [10-agent-api-guide.md](10-agent-api-guide.md) | End-to-end API guide for agents pricing servers from a TZ |
 | [11-lot-suggestions.md](11-lot-suggestions.md) | lot_suggestion column in qt_vendor_partnumber_seen — write/read contract for manual UI mappings |
+| [decisions/](decisions/README.md) | Architectural decision log, one file per decision |
 
 ## Rules
 
diff --git a/bible-local/decisions/2026-07-10-world-pricelist-fallback.md b/bible-local/decisions/2026-07-10-world-pricelist-fallback.md
index 97a68b4..23dc8ec 100644
--- a/bible-local/decisions/2026-07-10-world-pricelist-fallback.md
+++ b/bible-local/decisions/2026-07-10-world-pricelist-fallback.md
@@ -43,3 +43,7 @@ is missing or ≤ 0:
   "latest active `world` pricelist," not a per-configuration pin.
 - If no `world` pricelist exists locally, behavior is unchanged (missing prices stay "—",
   no error).
+- How the footer totals surface this fallback is decided separately in
+  [2026-07-24-pricing-total-world-share.md](2026-07-24-pricing-total-world-share.md);
+  membership of world LOTs in the configurator's LOT set is decided in
+  [2026-07-24-component-universe-world-union.md](2026-07-24-component-universe-world-union.md).
diff --git a/bible-local/decisions/2026-07-24-pricing-total-world-share.md b/bible-local/decisions/2026-07-24-pricing-total-world-share.md
new file mode 100644
index 0000000..c6735ae
--- /dev/null
+++ b/bible-local/decisions/2026-07-24-pricing-total-world-share.md
@@ -0,0 +1,47 @@
+# Decision: pricing totals signal the world-fallback share, not row coverage
+
+**Date:** 2026-07-24
+**Status:** active
+
+## Context
+
+The footer totals of both "Ценообразование" tables used to mean one thing: a red number
+with a trailing `*` said "prices are missing for some positions (N of M)". Two problems:
+
+- the asterisk was a separate `<span>` after the number and wrapped onto its own line,
+  pushing the footer layout around;
+- coverage was the wrong thing to warn about. Since
+  [2026-07-10-world-pricelist-fallback.md](2026-07-10-world-pricelist-fallback.md) a
+  missing price is silently filled from the `world` pricelist, so the total is rarely
+  incomplete — it is *contaminated*. A sum resting on world stand-in prices looked exactly
+  like a sum of real ones.
+
+## Decision
+
+`_setPricingTotal` in `web/templates/index.html` renders every footer total for Estimate,
+Склад and Конкуренты in both tables (buy and sale):
+
+- No asterisk, ever. The number stands alone.
+- The total is red **if and only if** part of it was computed from `world` prices — that
+  is, `worldTotal > 0`. A clean sum is not coloured. Red is a contamination signal, not a
+  coverage signal.
+- The world share per column is accumulated from the same per-row flags that tint the
+  cells amber (`estWorld` / `whWorld` / `compWorld`), summed over line totals, not unit
+  prices.
+- Hovering the total opens a styled popup (`.pricing-total-tip` in `web/static/app.css`),
+  not a `title=` tooltip — the native one appears too late to notice on a number the user
+  is meant to distrust. It states the exact percentage and, when prices are genuinely
+  missing for some rows, adds the coverage line that the asterisk used to carry.
+- "Ручная цена" is not part of this: it is user input, never world-derived.
+
+## Consequences
+
+- Red in the pricing footer means "part of this sum is a stand-in price", nothing else.
+  Do not reuse it for coverage, staleness or any other condition.
+- The colour and popup classes are re-applied on every render and stripped first, so they
+  never accumulate across re-renders.
+- The totals are written with `innerHTML` (number plus popup element). Nothing may read
+  `textContent` off `#pricing-total-*-estimate|warehouse|competitor`; the vendor-price
+  column and row `data-` attributes remain the read path for `applyCustomPrice` and
+  `_getPricingEstimateTotal`.
+- The old `_setPartialTotal` helper is gone; do not reintroduce a coverage-only signal.