Reject vision matches with action phase mismatches

cli.py

@@ -632,6 +632,66 @@ def _merge_best_results(existing: list, candidates: list, cfg) -> list:
     return sorted(by_id.values(), key=lambda r: r.beat_id)
 
 
+def _filter_semantically_invalid_vision_matches(results: list, beats: list, cfg) -> list:
+    """Drop vision-enabled matches whose final action phase contradicts the beat."""
+    if not cfg.vision.enabled or not results:
+        return results
+
+    from dataclasses import replace
+    from src.llm.vision_cache import validate_match_window_with_vision
+
+    logger = logging.getLogger(__name__)
+    beats_by_id = {beat.beat_id: beat for beat in beats}
+    kept = []
+    for result in results:
+        beat = beats_by_id.get(result.beat_id)
+        if beat is None:
+            kept.append(result)
+            continue
+
+        windows = []
+        if getattr(result, "segments", ()):
+            for segment in result.segments:
+                segment_beat = replace(
+                    beat,
+                    start_s=beat.start_s + segment.trailer_offset_s,
+                    end_s=beat.start_s + segment.trailer_offset_s + segment.duration_s,
+                )
+                windows.append((
+                    segment_beat,
+                    segment.scene_id,
+                    segment.in_point_s,
+                    segment.out_point_s,
+                ))
+        else:
+            windows.append((beat, result.scene_id, result.in_point_s, result.out_point_s))
+
+        valid = True
+        reasons: list[str] = []
+        for check_beat, scene_id, in_point_s, out_point_s in windows:
+            ok, reason = validate_match_window_with_vision(
+                check_beat,
+                source_path=result.source_path,
+                scene_id=scene_id,
+                in_point_s=in_point_s,
+                out_point_s=out_point_s,
+                cfg=cfg,
+            )
+            reasons.append(reason)
+            if not ok:
+                valid = False
+                break
+        if valid:
+            kept.append(result)
+        else:
+            logger.warning(
+                "Beat %d: rejected by vision action-phase verification (%s)",
+                result.beat_id,
+                "; ".join(reasons),
+            )
+    return kept
+
+
 def _attach_visual_segments(results: list, beats: list, cfg) -> list:
     """Attach automatic sub-shot matches for multi-island trailer beats."""
     from dataclasses import replace
@@ -976,6 +1036,7 @@ def cmd_match(args: argparse.Namespace, cfg) -> list:
         skip_global_segment_scan_for=set(single_island_trims),
     )
     results = _attach_visual_segments(results, beats, cfg)
+    results = _filter_semantically_invalid_vision_matches(results, beats, cfg)
 
     # A targeted one-beat match should improve the cache without deleting
     # automatic matches for other beats.