diff --git a/docs/superpowers/specs/2026-09-09-guided-local-import-design.md b/docs/superpowers/specs/2026-09-09-guided-local-import-design.md
index 4a68e8c..c12bd2e 100644
--- a/docs/superpowers/specs/2026-09-09-guided-local-import-design.md
+++ b/docs/superpowers/specs/2026-09-09-guided-local-import-design.md
@@ -56,17 +56,17 @@ Each scanned file has one of these statuses:
 
 - **Ready:** every candidate can be created without a validation issue or
   unresolved duplicate.
-- **Already imported:** every source candidate is already attached through
-  existing provenance; the file is skipped without financial writes.
+- **Already imported:** every candidate is already attached through existing provenance or
+  exactly matches a persisted transaction signature; the file is skipped without financial writes.
 - **Needs attention:** the file has an unsupported/unrecognized profile,
   malformed or unresolved candidate, missing account mapping, ambiguous
   potential duplicate, or missing/invalid statement context.
 
 The import action creates a normal existing preview per ready file. It applies
-`CREATE` to clean candidates. Exact source reimports receive
-`SKIP` with `ALREADY_IMPORTED`; they produce no transaction and are not an
-exception. Any potential duplicate that is not an exact existing source record
-requires review and prevents that file from committing automatically.
+`CREATE` to clean candidates. Exact source reimports and candidates with an exact persisted
+transaction signature receive `SKIP` with `ALREADY_IMPORTED`; they produce no transaction and
+are not an exception. Duplicates within the incoming file still require review and prevent that
+file from committing automatically.
 
 An exception in one file does not prevent other ready files from committing.
 Every file retains the import service's atomic commit boundary; a failed file
diff --git a/ingestion/guided_imports.py b/ingestion/guided_imports.py
index 26c3b0b..5e7c13c 100644
--- a/ingestion/guided_imports.py
+++ b/ingestion/guided_imports.py
@@ -374,7 +374,9 @@ def prepare_import(*, discovered: DiscoveredFile, section_mapping: dict[str, str
         decisions = []
         counts = {"created": 0, "matched": 0, "skipped": 0}
         for item in batch.items.order_by("locator"):
-            if item.issues == ["ALREADY_IMPORTED"]:
+            if item.issues == ["ALREADY_IMPORTED"] or (
+                not item.issues and item.duplicates["existing"] and not item.duplicates["within"]
+            ):
                 decisions.append({"locator": item.locator, "action": "SKIP", "reason": "ALREADY_IMPORTED"})
                 counts["skipped"] += 1
             elif not item.issues and not item.duplicates["existing"] and not item.duplicates["within"]:
diff --git a/tests/test_guided_import_workflow.py b/tests/test_guided_import_workflow.py
index e345718..eaf2d7a 100644
--- a/tests/test_guided_import_workflow.py
+++ b/tests/test_guided_import_workflow.py
@@ -83,14 +83,18 @@ def test_exact_second_import_only_skips_and_commit_is_idempotent(source, bank):
     assert_no_open_preview()
 
 
-def test_ambiguous_existing_duplicate_cancels_whole_file(source, bank):
+def test_existing_signature_duplicate_skips_without_creating_transactions(source, bank):
     commit(prepare(source(), bank))
-    ambiguous = prepare(source(DATA.replace(b",1,", b",3,")), bank)
-    assert ambiguous.state == "NEEDS_ATTENTION"
-    assert ambiguous.code == "REVIEW_REQUIRED"
-    assert Transaction.objects.count() == 2
-    assert_no_open_preview()
 
+    duplicate = prepare(source(DATA.replace(b",1,", b",3,")), bank)
+
+    assert duplicate.state == "ALREADY_IMPORTED"
+    assert duplicate.counts == {"created": 0, "matched": 0, "skipped": 2}
+    assert all(item.decision == {"action": "SKIP", "reason": "ALREADY_IMPORTED"}
+               for item in ImportBatch.objects.get(pk=duplicate.batch_id).items.all())
+    assert commit(duplicate).counts == {"created": 0, "matched": 0, "skipped": 2}
+    assert Transaction.objects.count() == SourceRecord.objects.count() == 2
+    assert_no_open_preview()
 
 @pytest.mark.parametrize("data", [
     (HEADER + ROW + ROW).encode(),