diff --git a/api/providers/vdb/vdb-tencent/src/dify_vdb_tencent/tencent_vector.py b/api/providers/vdb/vdb-tencent/src/dify_vdb_tencent/tencent_vector.py
index 30974156ab9352..d1f2491ec8a0a1 100644
--- a/api/providers/vdb/vdb-tencent/src/dify_vdb_tencent/tencent_vector.py
+++ b/api/providers/vdb/vdb-tencent/src/dify_vdb_tencent/tencent_vector.py
@@ -1,9 +1,9 @@
 import json
 import logging
 import math
-from typing import Any, TypedDict, override
+from typing import Any, TypedDict, cast, override
 
-from pydantic import BaseModel
+from pydantic import BaseModel, model_validator
 from sqlalchemy.orm import Session
 from tcvdb_text.encoder import BM25Encoder  # type: ignore
 from tcvectordb import RPCVectorDBClient, VectorDBException  # type: ignore
@@ -36,7 +36,7 @@ class TencentConfig(BaseModel):
     api_key: str | None = None
     timeout: float = 30
     username: str | None = None
-    database: str | None = None
+    database: str
     index_type: str = "HNSW"
     metric_type: str = "IP"
     shard: int = 1
@@ -44,6 +44,13 @@ class TencentConfig(BaseModel):
     max_upsert_batch_size: int = 128
     enable_hybrid_search: bool = False  # Flag to enable hybrid search
 
+    @model_validator(mode="before")
+    @classmethod
+    def validate_config(cls, values: dict[str, Any]) -> dict[str, Any]:
+        if not values.get("database"):
+            raise ValueError("config TENCENT_VECTOR_DB_DATABASE is required")
+        return values
+
     def to_tencent_params(self) -> TencentParamsDict:
         result: TencentParamsDict = {
             "url": self.url,
@@ -87,7 +94,10 @@ def _load_collection(self):
                     if idx.name == "sparse_vector":
                         has_hybrid_search = True
                     elif idx.name == "vector":
-                        self._dimension = idx.dimension
+                        # The "vector" index is the VectorIndex _create_collection made.
+                        dimension = cast(vdb_index.VectorIndex, idx).dimension
+                        if dimension is not None:
+                            self._dimension = dimension
                 if not has_hybrid_search:
                     self._enable_hybrid_search = False
 
@@ -151,7 +161,7 @@ def _create_collection(self, dimension: int):
                 index_type=enum.IndexType.SPARSE_INVERTED,
                 metric_type=enum.MetricType.IP,
             )
-            indexes = [index_id, index_vector, index_metadate]
+            indexes: list[vdb_index.IndexField] = [index_id, index_vector, index_metadate]
             if self._enable_hybrid_search:
                 indexes.append(index_sparse_vector)
             try:
@@ -170,7 +180,7 @@ def _create_collection(self, dimension: int):
                 index_metadate = vdb_index.FilterIndex(
                     self.field_metadata, enum.FieldType.String, enum.IndexType.FILTER
                 )
-                indexes = [index_id, index_vector, index_metadate]
+                indexes: list[vdb_index.IndexField] = [index_id, index_vector, index_metadate]
                 if self._enable_hybrid_search:
                     indexes.append(index_sparse_vector)
                 self._client.create_collection(
@@ -196,7 +206,7 @@ def add_texts(self, documents: list[Document], embeddings: list[list[float]], **
         batch_size = self._client_config.max_upsert_batch_size
         batch = math.ceil(total_count / batch_size)
         for j in range(batch):
-            docs = []
+            docs: list[document.Document | dict[str, Any]] = []
             start_idx = j * batch_size
             end_idx = min(total_count, (j + 1) * batch_size)
             for i in range(start_idx, end_idx):
@@ -267,7 +277,8 @@ def search_by_vector(self, query_vector: list[float], **kwargs: Any) -> list[Doc
             database_name=self._client_config.database,
             collection_name=self.collection_name,
             vectors=[query_vector],
-            filter=filter,
+            # tcvectordb declares `filter: Union[Filter, str] = None`; None means no filter.
+            filter=filter,  # pyrefly: ignore[bad-argument-type]
             params=document.HNSWSearchParams(ef=kwargs.get("ef", 10)),
             retrieve_vector=False,
             limit=kwargs.get("top_k", 4),
@@ -296,7 +307,8 @@ def search_by_full_text(self, query: str, **kwargs: Any) -> list[Document]:
             match=[
                 KeywordSearch(
                     field_name="sparse_vector",
-                    data=bm25.encode_queries(query),
+                    # A single query string encodes to a single sparse vector.
+                    data=cast(list[list[float | int]], bm25.encode_queries(query)),
                 ),
             ],
             rerank=WeightedRerank(
@@ -305,7 +317,8 @@ def search_by_full_text(self, query: str, **kwargs: Any) -> list[Document]:
             ),
             retrieve_vector=False,
             limit=kwargs.get("top_k", 4),
-            filter=filter,
+            # tcvectordb declares `filter: Union[Filter, str] = None`; None means no filter.
+            filter=filter,  # pyrefly: ignore[bad-argument-type]
         )
         score_threshold = float(kwargs.get("score_threshold") or 0.0)
         return self._get_search_res(res, score_threshold)
@@ -354,7 +367,7 @@ def init_vector(
                 api_key=dify_config.TENCENT_VECTOR_DB_API_KEY,
                 timeout=dify_config.TENCENT_VECTOR_DB_TIMEOUT,
                 username=dify_config.TENCENT_VECTOR_DB_USERNAME,
-                database=dify_config.TENCENT_VECTOR_DB_DATABASE,
+                database=dify_config.TENCENT_VECTOR_DB_DATABASE or "",
                 shard=dify_config.TENCENT_VECTOR_DB_SHARD,
                 replicas=dify_config.TENCENT_VECTOR_DB_REPLICAS,
                 enable_hybrid_search=dify_config.TENCENT_VECTOR_DB_ENABLE_HYBRID_SEARCH or False,
diff --git a/api/providers/vdb/vdb-tencent/tests/unit_tests/test_tencent_vector.py b/api/providers/vdb/vdb-tencent/tests/unit_tests/test_tencent_vector.py
index 5d4d05aec28bcd..259d058297731d 100644
--- a/api/providers/vdb/vdb-tencent/tests/unit_tests/test_tencent_vector.py
+++ b/api/providers/vdb/vdb-tencent/tests/unit_tests/test_tencent_vector.py
@@ -96,35 +96,36 @@ class _SparseIndex:
         def __init__(self, **kwargs):
             self.kwargs = kwargs
 
-    tcvectordb_enum.IndexType = SimpleNamespace(
+    # ModuleType does not declare these attributes, so set them through the module namespace.
+    vars(tcvectordb_enum)["IndexType"] = SimpleNamespace(
         __members__={"HNSW": "HNSW", "PRIMARY_KEY": "PRIMARY_KEY", "FILTER": "FILTER", "SPARSE_INVERTED": "SPARSE"},
         PRIMARY_KEY="PRIMARY_KEY",
         FILTER="FILTER",
         SPARSE_INVERTED="SPARSE",
     )
-    tcvectordb_enum.MetricType = SimpleNamespace(__members__={"IP": "IP"}, IP="IP")
-    tcvectordb_enum.FieldType = SimpleNamespace(String="String", Json="Json", SparseVector="SparseVector")
+    vars(tcvectordb_enum)["MetricType"] = SimpleNamespace(__members__={"IP": "IP"}, IP="IP")
+    vars(tcvectordb_enum)["FieldType"] = SimpleNamespace(String="String", Json="Json", SparseVector="SparseVector")
 
-    tcvectordb_document.Document = _Document
-    tcvectordb_document.HNSWSearchParams = _HNSWSearchParams
-    tcvectordb_document.AnnSearch = _AnnSearch
-    tcvectordb_document.Filter = _Filter
-    tcvectordb_document.KeywordSearch = _KeywordSearch
-    tcvectordb_document.WeightedRerank = _WeightedRerank
+    vars(tcvectordb_document)["Document"] = _Document
+    vars(tcvectordb_document)["HNSWSearchParams"] = _HNSWSearchParams
+    vars(tcvectordb_document)["AnnSearch"] = _AnnSearch
+    vars(tcvectordb_document)["Filter"] = _Filter
+    vars(tcvectordb_document)["KeywordSearch"] = _KeywordSearch
+    vars(tcvectordb_document)["WeightedRerank"] = _WeightedRerank
 
-    tcvectordb_index.HNSWParams = _HNSWParams
-    tcvectordb_index.FilterIndex = _FilterIndex
-    tcvectordb_index.VectorIndex = _VectorIndex
-    tcvectordb_index.SparseIndex = _SparseIndex
+    vars(tcvectordb_index)["HNSWParams"] = _HNSWParams
+    vars(tcvectordb_index)["FilterIndex"] = _FilterIndex
+    vars(tcvectordb_index)["VectorIndex"] = _VectorIndex
+    vars(tcvectordb_index)["SparseIndex"] = _SparseIndex
 
-    tcvdb_text_encoder.BM25Encoder = _BM25Encoder
+    vars(tcvdb_text_encoder)["BM25Encoder"] = _BM25Encoder
 
-    tcvectordb_model.document = tcvectordb_document
-    tcvectordb_model.enum = tcvectordb_enum
-    tcvectordb_model.index = tcvectordb_index
+    vars(tcvectordb_model)["document"] = tcvectordb_document
+    vars(tcvectordb_model)["enum"] = tcvectordb_enum
+    vars(tcvectordb_model)["index"] = tcvectordb_index
 
-    tcvectordb.RPCVectorDBClient = RPCVectorDBClient
-    tcvectordb.VectorDBException = VectorDBError
+    vars(tcvectordb)["RPCVectorDBClient"] = RPCVectorDBClient
+    vars(tcvectordb)["VectorDBException"] = VectorDBError
 
     return {
         "tcvdb_text": tcvdb_text,
@@ -186,7 +187,14 @@ def test_config_and_init_paths(tencent_module):
         indexes=[SimpleNamespace(name="vector", dimension=512)]
     )
     vector._load_collection()
-    assert vector._enable_hybrid_search is False
+    # Without a sparse_vector index, hybrid search is turned off.
+    assert not vector._enable_hybrid_search
+
+
+@pytest.mark.parametrize("database", [None, ""])
+def test_config_requires_a_database(tencent_module, database):
+    with pytest.raises(ValueError, match="config TENCENT_VECTOR_DB_DATABASE is required"):
+        _config(tencent_module, database=database)
 
 
 def test_create_collection_branches(tencent_module, monkeypatch: pytest.MonkeyPatch):
diff --git a/api/pyrefly-local-excludes.txt b/api/pyrefly-local-excludes.txt
index 8009889baf7277..e127244b8c3955 100644
--- a/api/pyrefly-local-excludes.txt
+++ b/api/pyrefly-local-excludes.txt
@@ -31,9 +31,7 @@ providers/vdb/vdb-qdrant/tests/unit_tests/test_qdrant_vector.py
 providers/vdb/vdb-relyt/tests/unit_tests/test_relyt_vector.py
 providers/vdb/vdb-tablestore/tests/integration_tests/test_tablestore.py
 providers/vdb/vdb-tablestore/tests/unit_tests/test_tablestore_vector.py
-providers/vdb/vdb-tencent/src/dify_vdb_tencent/tencent_vector.py
 providers/vdb/vdb-tencent/tests/integration_tests/conftest.py
-providers/vdb/vdb-tencent/tests/unit_tests/test_tencent_vector.py
 providers/vdb/vdb-tidb-on-qdrant/src/dify_vdb_tidb_on_qdrant/tidb_on_qdrant_vector.py
 providers/vdb/vdb-tidb-on-qdrant/src/dify_vdb_tidb_on_qdrant/tidb_service.py
 providers/vdb/vdb-tidb-on-qdrant/tests/unit_tests/test_tidb_service.py