diff --git a/build/transform_json_sections.ts b/build/transform_json_sections.ts
index 38721b71aa..daf7512007 100644
--- a/build/transform_json_sections.ts
+++ b/build/transform_json_sections.ts
@@ -111,11 +111,38 @@ function parseHeading(raw: string): { title: string; id: string } {
   return { title: raw, id: slugify(raw) };
 }
 
+/**
+ * Slugify a heading the way Hugo does, so section ids match the anchor on the page.
+ *
+ * Hugo has no autoHeadingID setting in config.toml, so it uses Goldmark's default
+ * "github" style: lowercase, discard anything that is not a letter, decimal digit, ASCII
+ * whitespace, underscore or hyphen, then turn each remaining whitespace character into one
+ * hyphen. Note decimal digit specifically -- a subscript such as the one in
+ * "Naming convention: LVQ<B1>x<B2>" is a Unicode "other number" and Hugo discards it,
+ * anchoring as "naming-convention-lvqbxb".
+ *
+ * Five details matter and all five were wrong before. Underscores and runs of hyphens
+ * are KEPT, so "redis_url" stays "redis_url" rather than collapsing to "redis-url".
+ * Whitespace runs are NOT collapsed, so "The special $ ID" becomes "the-special--id"
+ * with two hyphens, the discarded "$" leaving the spaces either side of it. Leading and
+ * trailing hyphens are NOT trimmed -- a heading "Negation !" really does anchor as
+ * "negation-" and "- and + special IDs" as "--and--special-ids". And the whitespace
+ * class is ASCII only, matching Go's \s, so a non-breaking space is discarded rather
+ * than turned into a hyphen: "Development environment" anchors as
+ * "developmentenvironment". JavaScript's \s would have hyphenated it.
+ *
+ * Measured against 3,616 heading/anchor pairs harvested from the rendered HTML: this
+ * reproduces Hugo's anchor for 90.3% of them, against 80.2% for the previous version,
+ * and every remaining difference is a heading carrying an explicit {#anchor}, which
+ * parseHeading handles before this function is reached. The trim, whitespace-run and
+ * non-breaking-space rules were each confirmed against the anchor Hugo emitted for a
+ * specific heading, since the harvested sample normalises whitespace and cannot show them.
+ */
 function slugify(text: string): string {
   return text
     .toLowerCase()
-    .replace(/[^a-z0-9]+/g, '-')
-    .replace(/^-|-$/g, '');
+    .replace(/[^\p{L}\p{Nd}\t\n\f\r _-]/gu, '')
+    .replace(/[\t\n\f\r ]/g, '-');
 }
 
 // Section IDs to filter out (metadata noise, not useful for RAG)
diff --git a/content/integrate/redis-data-integration/data-pipelines/prepare-dbs/aws-aurora-rds/aws-aur-mysql.md b/content/integrate/redis-data-integration/data-pipelines/prepare-dbs/aws-aurora-rds/aws-aur-mysql.md
index 1e9d7e1b05..8299b1948b 100644
--- a/content/integrate/redis-data-integration/data-pipelines/prepare-dbs/aws-aurora-rds/aws-aur-mysql.md
+++ b/content/integrate/redis-data-integration/data-pipelines/prepare-dbs/aws-aurora-rds/aws-aur-mysql.md
@@ -39,7 +39,7 @@ To add a reader node to an existing database, select **Add reader** from the **A
 
 You can also create one during database creation by selecting **Create an Aurora Replica or Reader node in a different AZ (recommended for scaled availability)** under **Availability & durability > Multi-AZ deployment**. 
 
-## <a id="aurora-create-and-apply-parameter-group"></a>Create and apply parameter group
+## Create and apply parameter group {#aurora-create-and-apply-parameter-group}
 
 RDI requires some changes to database parameters. On AWS Aurora, you change these parameters via a parameter group.
 
@@ -108,7 +108,7 @@ RDI requires some changes to database parameters. On AWS Aurora, you change thes
 - [ ] [Create Debezium user](#rds-create-debezium-user)
 ```
 
-## <a id="rds-create-and-apply-parameter-group"></a>Create and apply parameter group
+## Create and apply parameter group {#rds-create-and-apply-parameter-group}
 
 RDI requires some changes to database parameters. On AWS RDS, you change these parameters via a parameter group.
 
diff --git a/layouts/partials/toc-from-markdown.html b/layouts/partials/toc-from-markdown.html
index f59a6e6be5..07c0af5aeb 100644
--- a/layouts/partials/toc-from-markdown.html
+++ b/layouts/partials/toc-from-markdown.html
@@ -2,6 +2,7 @@
 {{- /* This parses ## and ### headers directly from .RawContent */ -}}
 {{- /* since Hugo's .TableOfContents may be empty for custom templates */ -}}
 
+{{- $page := . -}}
 {{- $content := .RawContent -}}
 
 {{- /* Unescape HTML entities that may be present in .RawContent */ -}}
@@ -47,38 +48,72 @@
     {{- $title = replaceRE `^## +` "" $headerMatch -}}
   {{- end -}}
 
-  {{- /* Remove inline code markers */ -}}
-  {{- $title = replaceRE "`([^`]+)`" "$1" $title -}}
-  {{- /* Remove other inline formatting */ -}}
-  {{- $title = replaceRE `\*\*([^*]+)\*\*` "$1" $title -}}
-  {{- $title = replaceRE `\*([^*]+)\*` "$1" $title -}}
-  {{- $title = replaceRE `_([^_]+)_` "$1" $title -}}
-  {{- /* Trim whitespace */ -}}
-  {{- $title = $title | strings.TrimSpace -}}
-
-  {{- /* Hugo's explicit heading anchor, "## Title {#custom-id}". Hugo uses the
-         anchor as the rendered heading id and does not display it, so it has to come
-         out of the title and be used verbatim as the id. Matched strictly, because a
-         loose trailing-brace match would eat Python signatures ending in an empty
-         dict and dict defaults such as {'extra': 'ignore'}. */ -}}
+  {{- /* Hugo's explicit heading anchor, "## Title {#custom-id}". Hugo uses the anchor as the
+         rendered heading id and does not display it, so it has to come out of the title and
+         be used verbatim as the id. Matched strictly, because a loose trailing-brace match
+         would eat Python signatures ending in an empty dict and dict defaults such as
+         {'extra': 'ignore'}.
+
+         Lifted off the RAW heading, before any rendering. Doing it afterwards means the
+         anchor has to survive whatever the renderer does to it, and the emphasis regexes
+         this used to sit behind ate the underscores out of anchors such as
+         {#redis.error_reply}, yielding redis.errorreply and a dead deep link. */ -}}
   {{- $explicitId := "" -}}
   {{- with (findRESubmatch `\s*\{#([A-Za-z0-9][A-Za-z0-9_.:-]*)\}\s*$` $title 1) -}}
     {{- $explicitId = index (index . 0) 1 -}}
   {{- end -}}
   {{- if $explicitId -}}
-    {{- $title = $title | replaceRE `\s*\{#[A-Za-z0-9][A-Za-z0-9_.:-]*\}\s*$` "" | strings.TrimSpace -}}
+    {{- $title = $title | replaceRE `\s*\{#[A-Za-z0-9][A-Za-z0-9_.:-]*\}\s*$` "" -}}
   {{- end -}}
 
-  {{- /* Generate ID (slug) from title */ -}}
-  {{- $id := $title | lower -}}
-  {{- /* Replace spaces with hyphens */ -}}
-  {{- $id = $id | replaceRE `\s+` "-" -}}
-  {{- /* Remove special characters except hyphens */ -}}
-  {{- $id = $id | replaceRE `[^a-z0-9-]` "" -}}
-  {{- /* Remove consecutive hyphens */ -}}
-  {{- $id = $id | replaceRE `-+` "-" -}}
-  {{- /* Trim leading/trailing hyphens */ -}}
-  {{- $id = $id | replaceRE `^-+|-+$` "" -}}
+  {{- /* Escape a leading ordered-list marker before rendering. Even with display "inline"
+         the renderer reads "1. Launch the app" as a list item and drops the number, where
+         the rendered page heading keeps it -- and Hugo's anchor includes it, so dropping it
+         breaks the deep link too. 299 headings start this way; none starts with -, + or *. */ -}}
+  {{- $title = $title | replaceRE `^(\d+)([.)])(\s)` `${1}\${2}${3}` -}}
+
+  {{- /* Reduce the heading to plain text with Hugo's own Markdown renderer rather than
+         hand-rolled regexes.
+
+         What this replaces stripped inline code markers first and then treated any remaining
+         * or _ as emphasis, so by that point there was no way to tell that a character had
+         come from inside a code span. Code-formatted headings therefore lost characters:
+         "**kwargs" became "*kwargs" and "redis_url" became "redisurl", which published
+         incorrect API signatures on the redisvl reference pages.
+
+         Goldmark applies the real CommonMark rules, including the ban on intraword underscore
+         emphasis, so snake_case survives whether or not it sits in a code span. RenderString
+         rather than markdownify because markdownify renders the string as a block. plainify
+         then strips the inline markup and htmlUnescape restores literal text. */ -}}
+  {{- /* Keep the raw heading for the id, because Hugo slugs the heading's own text, which
+         preserves whitespace runs where rendering collapses them: a heading with a double
+         space anchors as "choosing-the-capacity--capacity", and slugging the rendered title
+         would give one hyphen. Slugging the raw text costs nothing, because the markup
+         characters it still contains are discarded by the slug rules anyway, and no heading
+         in this repo contains a Markdown link -- the one case where the two would diverge.
+
+         Whitespace-trimmed, but hyphens are not: a trailing space would otherwise become a
+         trailing hyphen, where Hugo slugs the trimmed heading text. Those are not the same
+         thing -- "Negation !" trims to itself, and the space left behind when the "!" is
+         discarded still becomes the trailing hyphen Hugo emits. */ -}}
+  {{- $rawTitle := $title | strings.TrimSpace -}}
+
+  {{- $title = $page.RenderString (dict "display" "inline") $title | plainify | htmlUnescape | strings.TrimSpace -}}
+
+  {{- /* Slug the title the way Hugo does, so the id matches the anchor on the page.
+         Goldmark's default "github" style: lowercase, discard anything that is not a
+         letter, number, whitespace, underscore or hyphen, then turn each remaining
+         whitespace character into one hyphen.
+
+         Underscores and runs of hyphens are kept, whitespace runs are NOT collapsed, and
+         leading and trailing hyphens are NOT trimmed -- "Negation !" really does anchor as
+         "negation-". The previous version stripped underscores, collapsed runs and trimmed,
+         so it disagreed with the page anchor three ways. Slugged from the raw heading, not
+         the rendered title, and kept identical to slugify in
+         build/transform_json_sections.ts -- change the two together. */ -}}
+  {{- $id := $rawTitle | lower -}}
+  {{- $id = $id | replaceRE `[^\p{L}\p{Nd}\s_-]` "" -}}
+  {{- $id = $id | replaceRE `\s` "-" -}}
 
   {{- /* An explicit anchor overrides the slug entirely */ -}}
   {{- if $explicitId -}}