Improve suggested notes: limit 8, threshold 0.45, show relevance scores

- Raise similarity threshold 0.30 → 0.45: only genuinely relevant notes
  shown; loosely-related notes no longer pad the sidebar
- Increase max suggested notes 3 → 8 (zero added compute — threshold is
  the real gate; the embedding call is fixed regardless of limit)
- semantic_search_notes now returns list[tuple[float, Note]] instead of
  list[Note] so scores propagate through context_meta to the frontend
- Keyword fallback notes carry score=null (no cosine similarity available)
- ChatView sidebar shows % badge on each suggested note:
  green ≥75%, amber 60–74%, muted <60%
  Hovering reveals the raw score in a tooltip

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

frontend/src/types/chat.ts

@@ -54,7 +54,7 @@ export interface ConversationDetail extends Conversation {
 export interface ContextMeta {
   context_note_id: number | null;
   context_note_title: string | null;
-  auto_notes: { id: number; title: string }[];
+  auto_notes: { id: number; title: string; score?: number | null }[];
 }
 
 export interface OllamaStatus {

frontend/src/views/ChatView.vue

@@ -38,7 +38,7 @@ const includedNoteIds = ref<Set<number>>(new Set());
 const includedNotes = ref<{ id: number; title: string }[]>([]);
 
 // Suggested notes — auto-found by search, not yet included
-const suggestedNotes = ref<{ id: number; title: string }[]>([]);
+const suggestedNotes = ref<{ id: number; title: string; score?: number | null }[]>([]);
 
 let prevConvId: number | null = null;
 
@@ -482,6 +482,16 @@ onUnmounted(() => {
                 <router-link :to="`/notes/${note.id}`" class="context-note-name">
                   {{ note.title }}
                 </router-link>
+                <span
+                  v-if="note.score != null"
+                  class="context-note-score"
+                  :class="{
+                    'score-high':   note.score >= 0.75,
+                    'score-medium': note.score >= 0.60 && note.score < 0.75,
+                    'score-low':    note.score < 0.60,
+                  }"
+                  :title="`Relevance score: ${note.score}`"
+                >{{ Math.round(note.score * 100) }}%</span>
                 <button class="context-note-add" @click="includeNote(note)" title="Add to context">+</button>
               </div>
             </template>
@@ -807,6 +817,18 @@ onUnmounted(() => {
 .context-note-suggested {
   opacity: 0.85;
 }
+.context-note-score {
+  font-size: 0.67rem;
+  font-weight: 600;
+  padding: 0.05rem 0.22rem;
+  border-radius: 3px;
+  flex-shrink: 0;
+  font-variant-numeric: tabular-nums;
+  letter-spacing: 0;
+}
+.score-high   { color: var(--color-success); background: color-mix(in srgb, var(--color-success) 12%, transparent); }
+.score-medium { color: #c98a00; background: rgba(201, 138, 0, 0.12); }
+.score-low    { color: var(--color-text-muted); background: var(--color-bg-secondary); }
 .context-note-remove {
   background: none;
   border: none;

src/fabledassistant/services/embeddings.py

@@ -21,7 +21,9 @@ logger = logging.getLogger(__name__)
 
 # Minimum cosine similarity to include a note in context results.
 # nomic-embed-text produces unit-normalized vectors, so range is [-1, 1].
-_SIMILARITY_THRESHOLD = 0.30
+# 0.45 keeps only genuinely relevant notes; lower values like 0.30 let in
+# loosely-related results that pad the sidebar without adding real value.
+_SIMILARITY_THRESHOLD = 0.45
 
 
 async def get_embedding(text: str, model: str | None = None) -> list[float]:
@@ -75,10 +77,12 @@ async def semantic_search_notes(
     user_id: int,
     query: str,
     exclude_ids: set[int] | None = None,
-    limit: int = 3,
-) -> list[Note]:
-    """Return up to *limit* notes most relevant to *query* using cosine similarity.
+    limit: int = 8,
+) -> list[tuple[float, Note]]:
+    """Return up to *limit* (score, note) pairs most relevant to *query*.
 
+    Scores are cosine similarities in [-1, 1]; only notes at or above
+    _SIMILARITY_THRESHOLD are returned, sorted highest-first.
     Returns an empty list if the embedding model is unavailable or on any error.
     """
     try:
@@ -114,7 +118,7 @@ async def semantic_search_notes(
             scored.append((sim, note))
 
     scored.sort(key=lambda x: x[0], reverse=True)
-    return [note for _, note in scored[:limit]]
+    return scored[:limit]
 
 
 async def backfill_note_embeddings() -> None:

src/fabledassistant/services/llm.py

@@ -444,30 +444,38 @@ async def build_context(
     if current_note_id:
         search_exclude.add(current_note_id)
 
-    found_notes = []
+    # (score, note) pairs — score is float for semantic results, None for keyword fallback.
+    found_scored: list[tuple[float | None, object]] = []
+
     # Try semantic search first; fall back to keyword search on failure / no results.
     try:
         from fabledassistant.services.embeddings import semantic_search_notes
-        found_notes = await semantic_search_notes(
-            user_id, user_message, exclude_ids=search_exclude or None, limit=3
-        )
+        for score, note in await semantic_search_notes(
+            user_id, user_message, exclude_ids=search_exclude or None, limit=8
+        ):
+            found_scored.append((score, note))
     except Exception:
         logger.warning("Semantic note search failed, falling back to keyword search", exc_info=True)
 
-    if not found_notes:
+    if not found_scored:
         keywords = _extract_keywords(user_message)
         if keywords:
             try:
-                found_notes = await search_notes_for_context(
-                    user_id, keywords, exclude_ids=search_exclude or None, limit=3
-                )
+                for note in await search_notes_for_context(
+                    user_id, keywords, exclude_ids=search_exclude or None, limit=8
+                ):
+                    found_scored.append((None, note))
             except Exception:
                 logger.warning("Failed to search notes for context", exc_info=True)
 
-    # Populate sidebar candidates (never auto-injected).
-    for n in found_notes:
-        context_meta["auto_notes"].append({"id": n.id, "title": n.title})
-    context_meta["auto_note_ids"] = [n.id for n in found_notes]
+    # Populate sidebar candidates (never auto-injected into system prompt).
+    for score, n in found_scored:
+        context_meta["auto_notes"].append({
+            "id": n.id,
+            "title": n.title,
+            "score": round(score, 2) if score is not None else None,
+        })
+    context_meta["auto_note_ids"] = [n.id for _, n in found_scored]
 
     # Inject explicitly included notes (user opted in via sidebar click).
     if include_note_ids: