feat(snippets): near-duplicate finder — surface the sets worth merging

#231's premise was unifying reusable things already scattered as one-offs.
The create gate PREVENTS a new duplicate and merge_snippets CURES one you
point it at, but nothing FOUND the duplicates already in the record —
someone had to notice them by hand, which is the exact failure the Drafter
exists to remove.

One indexed self-join over note_embeddings, not an N² Python scan:
pgvector's cosine distance is the same operator semantic search uses, so a
similarity floor is a distance ceiling and the work stays in Postgres.
`left.note_id < right.note_id` yields each unordered pair once and drops
the self-pair that would otherwise dominate the ranking.

Pairs are collapsed into merge SETS by connected components. Transitive on
purpose: A~B plus B~C puts all three together even when A and C don't
directly clear the bar, which is what merge actually does (it folds every
source into one survivor). The cost is that a chain of mild resemblances
can rope in a member that isn't really alike — so the UI presents a set as
a proposal, shows the members, and never merges without a confirm.

Two scope decisions worth naming:

- OWN snippets only. merge_snippets requires one owner across the set, so
  surfacing someone else's would propose a merge that cannot be performed.
  The report is bounded by what the operator can act on, not what they can
  see.

- Threshold defaults to 0.82, LOOSER than the write gate's 0.90, and is a
  setting rather than a constant (rule #25). The gate blocks a create and
  has to be unforgiving of noise; this only suggests a merge under review,
  so it must reach further or it would never surface the pairs the gate
  already let through — which are precisely the ones that accumulated.

Fixes a real bug in the merge flow while wiring the UI: selectedList
filtered the selection against the CURRENT PAGE, and doMerge derives its
source ids from that list. A corpus-wide suggested group with off-page
members would have rendered incomplete and silently merged only the
visible subset. A group under review is now the authority for that list.

Refs #2088

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UaYUaouG9jjhATyuxCKrQs

src/scribe/mcp/server.py

@@ -256,6 +256,9 @@ _READ_ONLY_TOOLS = frozenset({
     "list_rules", "list_tags", "list_tasks", "list_topics", "list_trash",
     "list_always_on_rules", "search",
     "get_system", "list_systems", "list_system_records",
+    # Reports on the snippet corpus. Reads only — the merge it suggests is a
+    # separate, explicitly-called write.
+    "find_duplicate_snippets",
 })
 
 

src/scribe/mcp/tools/snippets.py

@@ -190,6 +190,44 @@ async def get_snippet(snippet_id: int) -> dict:
     return data
 
 
+async def find_duplicate_snippets(threshold: float = 0.0) -> dict:
+    """Find snippets already recorded that look like duplicates of each other.
+
+    The create gate PREVENTS a new duplicate and merge_snippets CURES one you
+    point it at — this is the missing third piece: it FINDS the ones already in
+    the record, so nobody has to notice them by hand.
+
+    Results are grouped into candidate merge SETS, not just pairs. Grouping is
+    transitive: if A resembles B and B resembles C, all three land in one set
+    even when A and C don't directly clear the bar. That mirrors what merge does
+    (it folds every source into one survivor), but it means a chain of mild
+    resemblances can rope in a member that isn't really alike — so read a set as
+    a proposal and check the members before acting.
+
+    Reports only YOUR snippets. merge_snippets requires one owner across the
+    whole set, so surfacing someone else's would propose a merge that can't be
+    performed.
+
+    Acting on a group: pick the best record as the canonical target, then
+    `merge_snippets(target_id, [other ids])`. Merge unions the fields and folds
+    every source's location in, so the survivor is findable at all their call
+    sites; the sources are trashed, recoverably. Prefer as target the one with
+    the clearest "when to reach for it" — merge keeps the target's title.
+
+    Args:
+        threshold: Similarity floor, 0-1. 0 (default) uses the configured
+            setting. Raise it if the report is noisy, lower it to catch more.
+
+    Returns {"groups": [{"note_ids", "snippets", "top_score"}], "pairs",
+    "threshold"}. An empty `groups` means nothing resembles anything else that
+    closely — the common and desirable case.
+    """
+    uid = current_user_id()
+    return await dedup_svc.find_duplicate_snippets(
+        uid, threshold=threshold if threshold > 0 else None
+    )
+
+
 async def verify_snippet(
     snippet_id: int, status: str, detail: str = "", path: str = "",
 ) -> dict:
@@ -365,6 +403,6 @@ async def merge_snippets(target_id: int, source_ids: list[int]) -> dict:
 def register(mcp) -> None:
     for fn in (
         list_snippets, create_snippet, get_snippet, update_snippet,
-        delete_snippet, merge_snippets, verify_snippet,
+        delete_snippet, merge_snippets, verify_snippet, find_duplicate_snippets,
     ):
         mcp.tool(name=fn.__name__)(fn)