alembic/versions/0023_note_link_target_id.py

@@ -0,0 +1,67 @@
+"""note_links.target_id — resolve [[links]] to a note, not to a string (M13 step 1)
+
+Revision ID: 0023
+Revises: 0022
+Create Date: 2026-08-22
+
+A wiki-link stored only as normalized TEXT means a note's name IS the edge: rename
+the note and every inbound link stops matching. The old answer was to rewrite the
+`[[Old Name]]` text inside every note that linked to it — workable while an explicit
+title existed to hold still, untenable once a note's name is just its first body
+line (M13).
+
+`target_norm` stays: it is what an UNRESOLVED link carries, since linking to a note
+that doesn't exist yet is a supported way to create one.
+
+The backfill is safe to run bluntly because note_links is DERIVED data — every row
+is recomputed from the source body on the next save regardless.
+"""
+from alembic import op
+import sqlalchemy as sa
+from sqlalchemy.dialects import postgresql
+
+revision = "0023"
+down_revision = "0022"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    op.add_column(
+        "note_links",
+        sa.Column("target_id", postgresql.UUID(as_uuid=True), nullable=True),
+    )
+    op.create_foreign_key(
+        "fk_note_links_target",
+        "note_links",
+        "notes",
+        ["target_id"],
+        ["id"],
+        # A deleted target un-resolves its inbound links rather than deleting them:
+        # the link text is still in the source's body, and it should read as pointing
+        # at something that isn't there — which is also what lets it re-resolve if a
+        # note of that name appears again.
+        ondelete="SET NULL",
+    )
+    op.create_index("ix_note_links_target_id", "note_links", ["target_id"])
+
+    # Resolve what can be resolved right now, scoped to the source's owner so a link
+    # can never bind to another user's note.
+    op.execute(
+        """
+        UPDATE note_links AS nl
+           SET target_id = t.id
+          FROM notes AS src, notes AS t
+         WHERE nl.source_id = src.id
+           AND t.owner_id = src.owner_id
+           AND t.deleted_at IS NULL
+           AND lower(btrim(t.display_title)) = nl.target_norm
+           AND t.id <> src.id
+        """
+    )
+
+
+def downgrade() -> None:
+    op.drop_index("ix_note_links_target_id", table_name="note_links")
+    op.drop_constraint("fk_note_links_target", "note_links", type_="foreignkey")
+    op.drop_column("note_links", "target_id")

frontend/src/components/MarkdownInline.vue

@@ -1,26 +1,49 @@
 <script setup lang="ts">
+import { computed } from "vue";
 import { useRouter } from "vue-router";
-import { useNotesStore } from "../stores/notes";
+import { useNotesStore, type NoteLinkRef } from "../stores/notes";
 import { useTitlesStore } from "../stores/titles";
 import type { InlineToken } from "../notes/markdown";
 
-defineProps<{ tokens: InlineToken[] }>();
+const props = defineProps<{ tokens: InlineToken[]; links?: NoteLinkRef[] }>();
 
 const router = useRouter();
 const notes = useNotesStore();
 const titles = useTitlesStore();
 
-// A [[wiki-link]] on the card: resolve the title and open the target note (creating
-// it first if it doesn't exist), via the board's ?open=<id> mechanism.
-async function follow(title: string) {
-  await titles.load();
-  let hit = titles.resolve(title);
-  if (!hit) {
-    const created = await notes.createTitled(title);
-    await titles.reload();
-    hit = { id: created.id, title: created.title ?? title };
+/**
+ * Where a [[link]] token actually points, according to the server.
+ *
+ * Keyed on the normalized written text, which is what survives in the body — the
+ * server resolved it to an id when the link was saved, so this keeps working after
+ * the target has been renamed and the written text has gone stale.
+ */
+const bound = computed(() => {
+  const map = new Map<string, NoteLinkRef>();
+  for (const l of props.links ?? []) map.set(l.norm, l);
+  return map;
+});
+
+/** What to SHOW for a link: the target's current name, else the text as written. */
+function label(token: string): string {
+  return bound.value.get(token.trim().toLowerCase())?.title ?? token;
+}
+
+// Open the target via the board's ?open=<id> mechanism, creating the note first if
+// the link names one that doesn't exist — which is a supported way to make a note.
+async function follow(token: string) {
+  const hit = bound.value.get(token.trim().toLowerCase());
+  if (hit) {
+    void router.push({ path: "/", query: { open: hit.id } });
+    return;
   }
-  void router.push({ path: "/", query: { open: hit.id } });
+  // No server binding: either this is running offline against the local store, or
+  // the link genuinely resolves to nothing. The name index answers the first case.
+  await titles.load();
+  const byName = titles.resolve(token);
+  const id = byName ? byName.id : (await notes.createTitled(token)).id;
+  if (!byName) await titles.reload();
+  void router.push({ path: "/", query: { open: id } });
 }
 </script>
 
@@ -35,7 +58,7 @@ async function follow(title: string) {
       class="cursor-pointer font-medium text-brand-700 underline-offset-2 hover:underline dark:text-brand"
       @click.stop="follow(t.value)"
       @keydown.enter.stop.prevent="follow(t.value)"
-      >{{ t.value }}</span
+      >{{ label(t.value) }}</span
     ><strong v-else-if="t.type === 'bold'" class="font-semibold">{{ t.value }}</strong
     ><em v-else-if="t.type === 'italic'">{{ t.value }}</em
     ><code

frontend/src/components/MarkdownText.vue

@@ -2,35 +2,38 @@
 import { computed } from "vue";
 import { parseMarkdown } from "../notes/markdown";
 import MarkdownInline from "./MarkdownInline.vue";
+import type { NoteLinkRef } from "../stores/notes";
 
-const props = defineProps<{ text: string }>();
+// `links` is the owning note's resolved [[links]], passed straight through to every
+// inline run — only MarkdownInline uses it, but only this component knows the note.
+const props = defineProps<{ text: string; links?: NoteLinkRef[] }>();
 const blocks = computed(() => parseMarkdown(props.text));
 </script>
 
 <template>
   <div class="space-y-1.5 break-words">
     <template v-for="(b, i) in blocks" :key="i">
-      <h3 v-if="b.type === 'h1'" class="text-base font-bold"><MarkdownInline :tokens="b.inline ?? []" /></h3>
-      <h4 v-else-if="b.type === 'h2'" class="text-sm font-bold"><MarkdownInline :tokens="b.inline ?? []" /></h4>
-      <h5 v-else-if="b.type === 'h3'" class="text-sm font-semibold"><MarkdownInline :tokens="b.inline ?? []" /></h5>
+      <h3 v-if="b.type === 'h1'" class="text-base font-bold"><MarkdownInline :tokens="b.inline ?? []" :links="links" /></h3>
+      <h4 v-else-if="b.type === 'h2'" class="text-sm font-bold"><MarkdownInline :tokens="b.inline ?? []" :links="links" /></h4>
+      <h5 v-else-if="b.type === 'h3'" class="text-sm font-semibold"><MarkdownInline :tokens="b.inline ?? []" :links="links" /></h5>
       <blockquote
         v-else-if="b.type === 'quote'"
         class="whitespace-pre-wrap border-l-2 border-neutral-300 pl-2 text-neutral-600 dark:border-neutral-600 dark:text-neutral-400"
       >
-        <MarkdownInline :tokens="b.inline ?? []" />
+        <MarkdownInline :tokens="b.inline ?? []" :links="links" />
       </blockquote>
       <ul v-else-if="b.type === 'ul'" class="list-disc space-y-0.5 pl-5">
-        <li v-for="(it, j) in b.items ?? []" :key="j"><MarkdownInline :tokens="it" /></li>
+        <li v-for="(it, j) in b.items ?? []" :key="j"><MarkdownInline :tokens="it" :links="links" /></li>
       </ul>
       <ol v-else-if="b.type === 'ol'" class="list-decimal space-y-0.5 pl-5">
-        <li v-for="(it, j) in b.items ?? []" :key="j"><MarkdownInline :tokens="it" /></li>
+        <li v-for="(it, j) in b.items ?? []" :key="j"><MarkdownInline :tokens="it" :links="links" /></li>
       </ol>
       <pre
         v-else-if="b.type === 'pre'"
         class="overflow-x-auto whitespace-pre-wrap rounded-md bg-black/5 p-2 font-mono text-xs dark:bg-white/10"
         >{{ b.value ?? "" }}</pre
       >
-      <p v-else class="whitespace-pre-wrap"><MarkdownInline :tokens="b.inline ?? []" /></p>
+      <p v-else class="whitespace-pre-wrap"><MarkdownInline :tokens="b.inline ?? []" :links="links" /></p>
     </template>
   </div>
 </template>

frontend/src/components/NoteCard.vue

@@ -237,7 +237,7 @@ onBeforeUnmount(() => document.removeEventListener("mousedown", onDocMousedown))
         {{ note.title }}
       </h3>
       <div v-if="note.body" class="text-sm text-neutral-700 dark:text-neutral-300">
-        <MarkdownText :text="note.body" />
+        <MarkdownText :text="note.body" :links="note.links" />
       </div>
       <p v-if="!note.title && !note.body && !note.attachments.length" class="text-sm italic text-neutral-400">
         Empty note

frontend/src/components/NoteEditor.vue

@@ -266,7 +266,13 @@ onMounted(async () => {
 });
 
 // ---- outgoing links (edit mode) ----
+//
+// Two sources, in order. The note's SAVED links carry the server's binding, so a
+// target that has since been renamed still resolves and is listed under its current
+// name. A link just typed into the textarea has no saved row yet, and the name index
+// is the best that can be said about it until the note is saved.
 const outgoingLinks = computed(() => {
+  const boundByNorm = new Map((props.note?.links ?? []).map((l) => [l.norm, l]));
   const re = /\[\[([^[\]]+)\]\]/g;
   const seen = new Set<string>();
   const out: { title: string; id: string | null }[] = [];
@@ -276,7 +282,8 @@ const outgoingLinks = computed(() => {
     const key = t.toLowerCase();
     if (t && !seen.has(key)) {
       seen.add(key);
-      out.push({ title: t, id: titles.resolve(t)?.id ?? null });
+      const hit = boundByNorm.get(key);
+      out.push(hit ? { title: hit.title, id: hit.id } : { title: t, id: titles.resolve(t)?.id ?? null });
     }
   }
   return out;

frontend/src/stores/notes.ts

@@ -63,6 +63,22 @@ export interface NoteRevision {
   created_at: string | null;
 }
 
+// One resolved [[wiki-link]] out of a note: the normalized text as WRITTEN, and the
+// note it actually points at with that note's name as it stands NOW.
+//
+// The server sends this because the client can no longer work it out. Resolution used
+// to be a name lookup in the titles index, which only held together because renaming
+// a note rewrote the link text inside every note that linked to it. Links are bound
+// by id now and bodies are left alone, so the written text can name something the
+// target is no longer called — and only the server holds the binding.
+export interface NoteLinkRef {
+  /** The link text as written, trimmed and lowercased — the key a token matches on. */
+  norm: string;
+  id: string;
+  /** The target's CURRENT name, which is what gets rendered. */
+  title: string;
+}
+
 export interface Note {
   id: string;
   title: string | null;
@@ -85,6 +101,11 @@ export interface Note {
   items: ChecklistItem[];
   attachments: Attachment[];
   previews: LinkPreview[];
+  // Absent offline: the desktop's local store derives links at query time and has no
+  // resolution to send. Rendering falls back to the titles index there, which is
+  // exactly right for a store where nothing else can have renamed the target behind
+  // this client's back.
+  links?: NoteLinkRef[];
   created_at: string | null;
   updated_at: string | null;
 }

src/thoughtsync/graph.py

@@ -1,7 +1,7 @@
 from __future__ import annotations
 
 from quart import Blueprint, g, jsonify
-from sqlalchemy import func, select
+from sqlalchemy import and_, func, or_, select
 from sqlalchemy.orm import aliased
 
 from .auth import login_required
@@ -33,11 +33,23 @@ async def get_graph():
     """
     source = aliased(Note)
     target = aliased(Note)
+    # A link joins on the note it was BOUND to, and only falls back to matching by
+    # name where it was never bound — a forward link written before its target
+    # existed. Name-matching alone is what used to make a rename break the graph.
     edge_stmt = (
         select(source.id, target.id)
         .select_from(NoteLink)
         .join(source, source.id == NoteLink.source_id)
-        .join(target, func.lower(func.trim(target.display_title)) == NoteLink.target_norm)
+        .join(
+            target,
+            or_(
+                target.id == NoteLink.target_id,
+                and_(
+                    NoteLink.target_id.is_(None),
+                    func.lower(func.trim(target.display_title)) == NoteLink.target_norm,
+                ),
+            ),
+        )
         .where(
             source.owner_id == g.user_id,
             source.deleted_at.is_(None),

src/thoughtsync/models/note_link.py

@@ -10,14 +10,38 @@ from . import Base
 
 
 class NoteLink(Base):
-    """A [[wiki-link]] from a source note to a target title (normalized). Resolved
-    to a target note by matching target_norm against lower(trim(note.title))."""
+    """A [[wiki-link]] from a source note to another note.
+
+    Two target columns, and the pair is the point:
+
+    - ``target_id`` — the note this link actually points at, bound when the link was
+      written. This is what makes a link survive its target being RENAMED. A note's
+      name is derived from its first body line, so without an id the name is the
+      edge, and editing that line would silently break every inbound link (or, in the
+      older design, force a rewrite of every linking note's body).
+    - ``target_norm`` — the normalized link text, always stored. It is what an
+      UNRESOLVED link carries: `[[a note that doesn't exist yet]]` is a supported way
+      to create one, so a link has to be able to name a target that isn't there.
+
+    Resolution reads the id first and falls back to matching the norm against
+    ``notes.display_title``, which is how a forward link connects the moment its
+    target appears. ``_claim_unresolved_links`` then binds the id, so the fallback is
+    a transitional state rather than a permanent one.
+    """
 
     __tablename__ = "note_links"
-    __table_args__ = (Index("ix_note_links_target", "target_norm"),)
+    __table_args__ = (
+        Index("ix_note_links_target", "target_norm"),
+        Index("ix_note_links_target_id", "target_id"),
+    )
 
     id: Mapped[uuid.UUID] = mapped_column(UUID(as_uuid=True), primary_key=True, default=uuid.uuid4)
     source_id: Mapped[uuid.UUID] = mapped_column(
         UUID(as_uuid=True), ForeignKey("notes.id", ondelete="CASCADE"), nullable=False
     )
+    # SET NULL rather than CASCADE: deleting the target un-resolves the link, it does
+    # not delete it. The link text is still sitting in the source's body.
+    target_id: Mapped[uuid.UUID | None] = mapped_column(
+        UUID(as_uuid=True), ForeignKey("notes.id", ondelete="SET NULL"), nullable=True
+    )
     target_norm: Mapped[str] = mapped_column(Text(), nullable=False)

src/thoughtsync/notes/__init__.py

@@ -19,7 +19,7 @@ import zipfile
 from datetime import datetime, timedelta, timezone
 
 from quart import Response, g, jsonify, request, send_file
-from sqlalchemy import case, func, literal_column, select
+from sqlalchemy import and_, case, func, literal_column, or_, select
 
 from ..acl import visible_to_user
 from ..auth import login_required
@@ -67,11 +67,10 @@ from .import_export import (
 )
 from .links import (
     _reconcile_tags,
-    _rename_inbound_links,
+    _claim_unresolved_links,
     _rewrite_links,
     parse_link_titles,
     parse_tags,
-    rewrite_link_title,
 )
 from .recurrence import REMINDER_RECURRENCES, next_occurrence, normalize_recurrence
 from .serialize import _items_for_notes, _labels_for_notes, _serialize_note, _serialize_notes
@@ -83,12 +82,11 @@ __all__ = [
     "parse_list_items",
     "parse_tags",
     "parse_link_titles",
-    "rewrite_link_title",
     "normalize_color",
     "normalize_recurrence",
     "next_occurrence",
     "_reconcile_tags",
-    "_rename_inbound_links",
+    "_claim_unresolved_links",
     "_rewrite_links",
     "_serialize_notes",
     "_escape_like",
@@ -435,15 +433,20 @@ async def note_backlinks(note_id: str):
         )
         if note is None:
             return not_found()
-        if not note.display_title:
-            return jsonify({"backlinks": []})
-        norm = note.display_title.strip().lower()
+        # Bound links (target_id) OR unbound ones still naming this note. The second
+        # half is what catches a link written before this note existed and not yet
+        # claimed; without it a forward link would go quiet until its source is next
+        # saved.
+        norm = (note.display_title or "").strip().lower()
+        matches = NoteLink.target_id == nid
+        if norm:
+            matches = or_(matches, and_(NoteLink.target_id.is_(None), NoteLink.target_norm == norm))
         sources = (
             await db.scalars(
                 select(Note)
                 .join(NoteLink, NoteLink.source_id == Note.id)
                 .where(
-                    NoteLink.target_norm == norm,
+                    matches,
                     Note.owner_id == g.user_id,
                     Note.deleted_at.is_(None),
                     Note.id != nid,
@@ -524,6 +527,9 @@ async def create_note():
             db.add(NoteItem(note_id=note.id, text=text, position=pos))
         await _rewrite_links(db, note)
         await _reconcile_tags(db, note)
+        # A new note may be exactly what earlier `[[links]]` were pointing at — the
+        # create-by-linking flow writes the link first and the note second.
+        await _claim_unresolved_links(db, note)
         await db.commit()
         await db.refresh(note)
         return jsonify(await _serialize_note(db, note)), 201
@@ -587,12 +593,12 @@ async def update_note(note_id: str):
         if "body" in data:
             await _rewrite_links(db, note)
             await _reconcile_tags(db, note)
-        # The display NAME changing — via an explicit title OR the first body line —
-        # repoints inbound [[Old Name]] references so backlinks survive (skip pure
-        # case/whitespace changes, which still resolve).
+        # A rename no longer touches anything else's TEXT. Inbound links already hold
+        # this note's id, so they follow it automatically; all that is left is to
+        # adopt any still-unresolved link that was waiting for this name.
         new_display = note.display_title
-        if old_display and new_display and old_display.strip().lower() != new_display.strip().lower():
-            await _rename_inbound_links(db, note, old_display, new_display)
+        if old_display != new_display:
+            await _claim_unresolved_links(db, note)
         # Version history: snapshot the PRE-edit title+body whenever either changed.
         if note.title != old_title or note.body != old_body:
             db.add(NoteRevision(note_id=note.id, title=old_title, body=old_body))
@@ -653,8 +659,8 @@ async def restore_revision(note_id: str, rev_id: str):
         await _rewrite_links(db, note)
         await _reconcile_tags(db, note)
         new_display = note.display_title
-        if old_display and new_display and old_display.strip().lower() != new_display.strip().lower():
-            await _rename_inbound_links(db, note, old_display, new_display)
+        if old_display != new_display:
+            await _claim_unresolved_links(db, note)
         await db.commit()
         await db.refresh(note)
         return jsonify(await _serialize_note(db, note))

src/thoughtsync/notes/links.py

@@ -5,7 +5,8 @@ from __future__ import annotations
 
 import re
 
-from sqlalchemy import delete, func, select
+from sqlalchemy import delete, func, select, update
+from sqlalchemy.orm import aliased
 
 from ..models.label import Label, NoteLabel
 from ..models.note import Note
@@ -49,11 +50,65 @@ def parse_link_titles(body: str | None) -> list[str]:
     return out
 
 
+async def _resolve_target(db, owner_id, norm: str, exclude_id=None):
+    """The owner's note currently NAMED `norm`, or None.
+
+    Owner-scoped so a link can never bind to someone else's note, and self-excluded
+    so a note that opens with its own name doesn't link to itself.
+    """
+    stmt = select(Note.id).where(
+        Note.owner_id == owner_id,
+        Note.deleted_at.is_(None),
+        func.lower(func.trim(Note.display_title)) == norm,
+    )
+    if exclude_id is not None:
+        stmt = stmt.where(Note.id != exclude_id)
+    return await db.scalar(stmt)
+
+
 async def _rewrite_links(db, note: Note) -> None:
-    """Replace a note's outgoing wiki-links from its current body."""
+    """Replace a note's outgoing wiki-links from its current body.
+
+    Each link is bound to the target's ID where one exists under that name right now.
+    That binding is what survives the target being renamed later; the norm is kept
+    either way, so a link to a note that doesn't exist yet is still recorded and can
+    resolve when it does.
+    """
     await db.execute(delete(NoteLink).where(NoteLink.source_id == note.id))
     for norm in parse_link_titles(note.body):
-        db.add(NoteLink(source_id=note.id, target_norm=norm))
+        target_id = await _resolve_target(db, note.owner_id, norm, exclude_id=note.id)
+        db.add(NoteLink(source_id=note.id, target_norm=norm, target_id=target_id))
+
+
+async def _claim_unresolved_links(db, note: Note) -> None:
+    """Bind still-unresolved links that name this note to it.
+
+    Called when a note's display name changes or a note is created. Two cases, one
+    mechanism: someone wrote `[[groceries]]` before any note was called that, or a
+    note has just been renamed INTO a name that other notes were already pointing at.
+
+    This is what replaced `_rename_inbound_links`, and the difference is the whole
+    point of the change: that function edited the BODIES of other people's notes to
+    keep their link text matching. This touches only link rows. A note's text is
+    never modified by something happening to a different note.
+    """
+    norm = (note.display_title or "").strip().lower()
+    if not norm:
+        return
+    source = aliased(Note)
+    unresolved = (
+        select(NoteLink.id)
+        .join(source, source.id == NoteLink.source_id)
+        .where(
+            NoteLink.target_id.is_(None),
+            NoteLink.target_norm == norm,
+            source.owner_id == note.owner_id,
+            source.id != note.id,
+        )
+    )
+    await db.execute(
+        update(NoteLink).where(NoteLink.id.in_(unresolved.scalar_subquery())).values(target_id=note.id)
+    )
 
 
 async def _find_or_create_label(db, owner_id, name: str):
@@ -89,38 +144,3 @@ async def _reconcile_tags(db, note: Note) -> None:
         if lid not in attached_ids:
             db.add(NoteLabel(note_id=note.id, label_id=lid, via_tag=True))
             attached_ids.add(lid)
-
-
-def rewrite_link_title(body: str | None, old_norm: str, new_title: str) -> str:
-    """Repoint every [[token]] whose normalized form == old_norm to [[new_title]]."""
-    if not body:
-        return body or ""
-
-    def _sub(match: re.Match) -> str:
-        return f"[[{new_title}]]" if match.group(1).strip().lower() == old_norm else match.group(0)
-
-    return _LINK_RE.sub(_sub, body)
-
-
-async def _rename_inbound_links(db, renamed: Note, old_title: str, new_title: str) -> None:
-    """Rewrite [[old title]] references (and their link rows) in every note that
-    links to the renamed note, so its backlinks survive the title change."""
-    old_norm = old_title.strip().lower()
-    sources = (
-        await db.scalars(
-            select(Note)
-            .join(NoteLink, NoteLink.source_id == Note.id)
-            .where(
-                NoteLink.target_norm == old_norm,
-                Note.owner_id == renamed.owner_id,
-                Note.deleted_at.is_(None),
-            )
-        )
-    ).all()
-    seen: set = set()
-    for source in sources:
-        if source.id in seen:
-            continue
-        seen.add(source.id)
-        source.body = rewrite_link_title(source.body, old_norm, new_title)
-        await _rewrite_links(db, source)

src/thoughtsync/notes/serialize.py

@@ -1,14 +1,16 @@
-"""Note serialization — turn a Note (+ its labels/items/attachments/previews) into
-the JSON dict the API returns. The bulk loaders (`*_for_notes`) fetch each child
+"""Note serialization — turn a Note (+ its labels/items/attachments/previews/links)
+into the JSON dict the API returns. The bulk loaders (`*_for_notes`) fetch each child
 collection for a batch of notes in one query, so list endpoints avoid N+1s."""
 from __future__ import annotations
 
-from sqlalchemy import select
+from sqlalchemy import and_, func, or_, select
+from sqlalchemy.orm import aliased
 
 from ..models.label import Label, NoteLabel
 from ..models.note import Note
 from ..models.note_attachment import NoteAttachment
 from ..models.note_item import NoteItem
+from ..models.note_link import NoteLink
 from ..models.note_link_preview import NoteLinkPreview
 
 
@@ -104,6 +106,64 @@ async def _previews_for_notes(db, note_ids: list) -> dict:
     return result
 
 
+async def _links_for_notes(db, note_ids: list) -> dict:
+    """Map note_id -> [{norm, id, title}] for each note's RESOLVED outgoing links.
+
+    The client cannot work this out for itself any more, and that is deliberate. It
+    used to resolve `[[text]]` by looking the text up in a client-side name index,
+    which only worked because a rename rewrote the text in every linking note. Now
+    that a link is bound to an id and the text is left alone, the stored text can name
+    something the target is no longer called — so the server, which holds the binding,
+    is the only place that knows where a link goes.
+
+    `title` is the target's name RIGHT NOW, so a renamed note reads correctly
+    everywhere it is linked from without a single body having been edited.
+
+    Unresolved links are simply absent: the client renders those as the
+    create-on-click affordance it already has.
+    """
+    if not note_ids:
+        return {}
+    source = aliased(Note)
+    target = aliased(Note)
+    rows = (
+        await db.execute(
+            select(NoteLink.source_id, NoteLink.target_norm, target.id, target.display_title)
+            .select_from(NoteLink)
+            .join(source, source.id == NoteLink.source_id)
+            .join(
+                target,
+                or_(
+                    target.id == NoteLink.target_id,
+                    and_(
+                        NoteLink.target_id.is_(None),
+                        func.lower(func.trim(target.display_title)) == NoteLink.target_norm,
+                    ),
+                ),
+            )
+            .where(
+                NoteLink.source_id.in_(note_ids),
+                target.deleted_at.is_(None),
+                # Owner-scoped, and NOT optional. A bound target_id was resolved
+                # owner-scoped when it was written, but the name fallback matches on
+                # display_title alone — without this, two users who both have a note
+                # called "Groceries" would leak each other's note id and name through
+                # an unresolved link. (Rule 47.)
+                target.owner_id == source.owner_id,
+            )
+        )
+    ).all()
+    result: dict = {}
+    for source_id, norm, target_id, title in rows:
+        bucket = result.setdefault(source_id, [])
+        # The name-fallback join can produce more than one candidate for the same
+        # text; first one wins, deterministically enough for a display hint.
+        if any(link["norm"] == norm for link in bucket):
+            continue
+        bucket.append({"norm": norm, "id": str(target_id), "title": title})
+    return result
+
+
 async def _serialize_note(db, note: Note) -> dict:
     data = note.serialize()
     labels = await _labels_for_notes(db, [note.id])
@@ -114,6 +174,8 @@ async def _serialize_note(db, note: Note) -> dict:
     data["attachments"] = attachments.get(note.id, [])
     previews = await _previews_for_notes(db, [note.id])
     data["previews"] = previews.get(note.id, [])
+    links = await _links_for_notes(db, [note.id])
+    data["links"] = links.get(note.id, [])
     return data
 
 
@@ -123,6 +185,7 @@ async def _serialize_notes(db, notes: list) -> list:
     items_map = await _items_for_notes(db, ids)
     attach_map = await _attachments_for_notes(db, ids)
     preview_map = await _previews_for_notes(db, ids)
+    link_map = await _links_for_notes(db, ids)
     out = []
     for n in notes:
         data = n.serialize()
@@ -130,5 +193,6 @@ async def _serialize_notes(db, notes: list) -> list:
         data["items"] = items_map.get(n.id, [])
         data["attachments"] = attach_map.get(n.id, [])
         data["previews"] = preview_map.get(n.id, [])
+        data["links"] = link_map.get(n.id, [])
         out.append(data)
     return out

src/thoughtsync/sync.py

@@ -28,7 +28,7 @@ from .models.note_item import NoteItem
 from .models.note_revision import NoteRevision
 from .notes import (
     _reconcile_tags,
-    _rename_inbound_links,
+    _claim_unresolved_links,
     _rewrite_links,
     _serialize_notes,
     derive_display_title,
@@ -295,9 +295,12 @@ async def _apply_note(db, ch: dict) -> dict:
     await _rewrite_links(db, note)
     await _reconcile_tags(db, note)
     await _apply_note_manual_labels(db, note, ch)
+    # Any change to the name — including a note arriving for the first time, where
+    # the old name was empty — may be what unresolved inbound links were waiting for.
+    # Nothing else's body is touched; see _claim_unresolved_links.
     new_display = note.display_title
-    if old_display and new_display and old_display.strip().lower() != new_display.strip().lower():
-        await _rename_inbound_links(db, note, old_display, new_display)
+    if old_display != new_display:
+        await _claim_unresolved_links(db, note)
     await db.flush()
     await db.refresh(note, ["sync_revision"])
     return {

tests/test_notes.py

@@ -22,7 +22,6 @@ from thoughtsync.notes import (
     parse_link_titles,
     parse_list_items,
     parse_tags,
-    rewrite_link_title,
 )
 
 
@@ -128,15 +127,22 @@ def test_parse_link_titles_empty():
     assert parse_link_titles("no links here") == []
 
 
-def test_rewrite_link_title():
-    body = "see [[Alpha]] and [[ alpha ]] and [[Beta]]"
-    assert rewrite_link_title(body, "alpha", "Gamma") == "see [[Gamma]] and [[Gamma]] and [[Beta]]"
+# `rewrite_link_title` and `_rename_inbound_links` are gone (M13 step 1). They kept
+# backlinks alive across a rename by editing the [[text]] inside every note that
+# linked to the renamed one — which meant one note's edit silently rewrote another's
+# words. Links are bound to a note id now, so a rename needs no repair at all.
+#
+# What replaced them (`_resolve_target`, `_claim_unresolved_links`, and the id-first
+# resolution in backlinks / the graph / serialization) is all SQL, and this suite runs
+# without a database, so it is deliberately not asserted here. See the task log: that
+# behaviour was checked by hand, and this repo has no integration lane to hold it.
 
 
-def test_rewrite_link_title_noop():
-    assert rewrite_link_title("", "alpha", "Gamma") == ""
-    assert rewrite_link_title(None, "alpha", "Gamma") == ""
-    assert rewrite_link_title("no links here", "alpha", "Gamma") == "no links here"
+def test_parse_link_titles_ignores_nesting():
+    # The link regex refuses [ and ] inside a token, so a malformed nest yields the
+    # inner name rather than something spanning both — worth pinning, since this is
+    # the string that becomes a link's stored target.
+    assert parse_link_titles("[[outer [[inner]] ]]") == ["inner"]
 
 
 def test_derive_display_title_explicit_wins():