diff --git a/.gitignore b/.gitignore
index 5e58801..eeeef7f 100644
--- a/.gitignore
+++ b/.gitignore
@@ -12,3 +12,4 @@ wheels/
 .env
 .claude
 .vscode
+.ruff_cache
diff --git a/lrcfetch/cache.py b/lrcfetch/cache.py
index a8b83ea..8c60df3 100644
--- a/lrcfetch/cache.py
+++ b/lrcfetch/cache.py
@@ -16,7 +16,9 @@ from .config import DB_PATH, DURATION_TOLERANCE_MS
 from .models import TrackMeta, LyricResult, CacheStatus
 
 # Punctuation to strip for fuzzy matching (ASCII + common fullwidth)
-_PUNCT_RE = re.compile(r"[~!@#$%^&*()_+\-=\[\]{}|;:'\",.<>?/\\`～！＠＃＄％＾＆＊（）＿＋－＝【】｛｝｜；：＇＂，。＜＞？／＼｀]")
+_PUNCT_RE = re.compile(
+    r"[~!@#$%^&*()_+\-=\[\]{}|;:'\",.<>?/\\`～！＠＃＄％＾＆＊（）＿＋－＝【】｛｝｜；：＇＂，。＜＞？／＼｀]"
+)
 _SPACE_RE = re.compile(r"\s+")
 
 
@@ -258,6 +260,45 @@ class CacheEngine:
             params.append(track.album)
         return conditions, params
 
+    # Exact cross-source search
+
+    def find_best_positive(self, track: TrackMeta) -> Optional[LyricResult]:
+        """Find the best positive (synced/unsynced) cache entry for *track*.
+
+        Uses exact metadata match (artist + title + album) across all sources.
+        Returns synced if available, otherwise unsynced, or None.
+        """
+        conditions, params = self._track_where(track)
+        if not conditions:
+            return None
+
+        now = int(time.time())
+        conditions.append("status IN (?, ?)")
+        params.extend(
+            [CacheStatus.SUCCESS_SYNCED.value, CacheStatus.SUCCESS_UNSYNCED.value]
+        )
+        conditions.append("(expires_at IS NULL OR expires_at > ?)")
+        params.append(str(now))
+
+        where = " AND ".join(conditions)
+        with sqlite3.connect(self.db_path) as conn:
+            conn.row_factory = sqlite3.Row
+            rows = conn.execute(
+                f"SELECT status, lyrics, source FROM cache WHERE {where} "
+                "ORDER BY CASE status WHEN ? THEN 0 ELSE 1 END LIMIT 1",
+                params + [CacheStatus.SUCCESS_SYNCED.value],
+            ).fetchall()
+
+        if not rows:
+            return None
+
+        row = dict(rows[0])
+        return LyricResult(
+            status=CacheStatus(row["status"]),
+            lyrics=row["lyrics"],
+            source="cache-search",
+        )
+
     # Fuzzy search
 
     def search_by_meta(
@@ -318,7 +359,12 @@ class CacheEngine:
                 else:
                     # No duration info in cache — still a candidate but lower priority
                     scored.append((DURATION_TOLERANCE_MS, m))
-            scored.sort(key=lambda x: (x[0], x[1].get("status") != CacheStatus.SUCCESS_SYNCED.value))
+            scored.sort(
+                key=lambda x: (
+                    x[0],
+                    x[1].get("status") != CacheStatus.SUCCESS_SYNCED.value,
+                )
+            )
             matches = [m for _, m in scored]
 
         return matches
diff --git a/lrcfetch/fetchers/cache_search.py b/lrcfetch/fetchers/cache_search.py
index 14fb0ae..7d784f8 100644
--- a/lrcfetch/fetchers/cache_search.py
+++ b/lrcfetch/fetchers/cache_search.py
@@ -31,6 +31,13 @@ class CacheSearchFetcher(BaseFetcher):
             logger.debug("Cache-search: skipped — no title")
             return None
 
+        # Fast path: exact metadata match (artist+title+album), single SQL query
+        exact = self._cache.find_best_positive(track)
+        if exact:
+            logger.info(f"Cache-search: exact hit ({exact.status.value})")
+            return exact
+
+        # Slow path: fuzzy cross-album search
         matches = self._cache.search_by_meta(
             artist=track.artist,
             title=track.title,
@@ -55,7 +62,7 @@ class CacheSearchFetcher(BaseFetcher):
 
         status = CacheStatus(best["status"])
         logger.info(
-            f"Cache-search: hit from [{best.get('source')}] "
+            f"Cache-search: fuzzy hit from [{best.get('source')}] "
             f"album={best.get('album')!r} ({status.value})"
         )
         return LyricResult(