diff --git a/src/rosetta-cli/rosetta_cli/cli.py b/src/rosetta-cli/rosetta_cli/cli.py
index 5dbb6c5e..16e8a0cf 100644
--- a/src/rosetta-cli/rosetta_cli/cli.py
+++ b/src/rosetta-cli/rosetta_cli/cli.py
@@ -378,7 +378,8 @@ def main() -> int:
             chunk_method=config.chunk_method,
             parser_config=config.parser_config,
             page_size=config.page_size,
-            timeout=config.timeout
+            dataset_default=config.dataset_default,
+            timeout=config.timeout,
         )
         
         return execute_command(args.command, args, client, config)
diff --git a/src/rosetta-cli/rosetta_cli/ragflow_client.py b/src/rosetta-cli/rosetta_cli/ragflow_client.py
index 9286211a..7efc7ab9 100644
--- a/src/rosetta-cli/rosetta_cli/ragflow_client.py
+++ b/src/rosetta-cli/rosetta_cli/ragflow_client.py
@@ -116,7 +116,8 @@ def __init__(
         embedding_model: str | None = None,
         chunk_method: str = "naive",
         parser_config: JsonDict | None = None,
-        page_size: int = 1000
+        page_size: int = 1000,
+        dataset_default: str = "aia",
     ):
         """
         Initialize RAGFlow client.
@@ -130,6 +131,7 @@ def __init__(
             chunk_method: Chunking method (default: naive)
             parser_config: Parser configuration dict for chunk_method settings
             page_size: Default page size for list operations (default: 1000)
+            dataset_default: Default dataset name when no release-specific or explicit dataset name is available
             
         Raises:
             ValueError: If api_key or base_url is empty
@@ -147,6 +149,7 @@ def __init__(
         self.chunk_method = chunk_method
         self.parser_config = parser_config or {}
         self.page_size = page_size
+        self.dataset_default = dataset_default
 
         # Initialize RAGFlow SDK client
         self._client = RAGFlow(api_key=api_key, base_url=base_url, version=version)
@@ -469,26 +472,38 @@ def _ensure_dataset(self, name: str, description: str = "", dry_run: bool = Fals
         # Dataset doesn't exist, create it (gated by dry_run)
         return self.create_dataset(name, description, dry_run=dry_run)
     
-    def _resolve_dataset_name(self, template: str, release: str | None) -> str:
+    def _resolve_dataset_name(
+        self,
+        dataset_template: str,
+        release: str | None,
+        dataset_name: str | None,
+    ) -> str:
         """
-        Resolve dataset name from template.
+        Resolve dataset name using release + dataset_template, explicit dataset_name or default.
         
         Args:
-            template: Name template (e.g., "aia-{release}")
+            dataset_template: Name template (e.g., "aia-{release}")
             release: Release identifier (e.g., "r1")
+            dataset_name: Explicit dataset name, if provided (e.g., "aia-r1")
             
         Returns:
             Resolved dataset name
             
         Examples:
-            >>> _resolve_dataset_name("aia-{release}", "r1")
+            >>> _resolve_dataset_name("aia-{release}", "r1", "aia-r2")
             "aia-r1"
-            >>> _resolve_dataset_name("aia", None)
-            "aia"
+            >>> _resolve_dataset_name("aia-{release}", None, "aia-r2")
+            "aia-r2"
         """
-        if release and "{release}" in template:
-            return template.format(release=release)
-        return template
+        if release and "{release}" in dataset_template:
+            resolved_name = dataset_template.format(release=release)
+        else:
+            resolved_name = dataset_name or self.dataset_default
+
+        if "{release}" in resolved_name:
+            raise ValueError(f"Unresolved release placeholder in dataset name: {resolved_name}")
+
+        return resolved_name
     
     def _build_title_with_tags(self, tags: list[str], filename: str) -> str:
         """
@@ -521,7 +536,7 @@ def upload_document(
         OPTIMIZED: Now accepts pre-read content to avoid redundant file I/O.
         
         This method:
-        1. Resolves dataset name from template + release
+        1. Resolves the target dataset name
         2. Ensures dataset exists
         3. Builds title with tag prefixes
         4. Checks if document exists (by ims_doc_id)
@@ -529,6 +544,12 @@ def upload_document(
         6. Deletes existing document if changed
         7. Uploads new document with metadata
         
+        Dataset name resolution precedence:
+        1. If metadata.release is present, and dataset_template contains "{release}",
+           resolve dataset_template with the release.
+        2. Otherwise, use dataset_name when provided.
+        3. Otherwise, fall back to dataset_default.
+
         Args:
             file_path: Path to file (for filename, backward compatibility)
             metadata: Document metadata with pre-calculated hash
@@ -567,10 +588,7 @@ def upload_document(
         actual_hash = metadata.content_hash
         
         # Resolve dataset name
-        resolved_name = self._resolve_dataset_name(
-            dataset_template if "{release}" in dataset_template else (dataset_name or dataset_template),
-            metadata.release
-        )
+        resolved_name = self._resolve_dataset_name(dataset_template, metadata.release, dataset_name)
         
         # Ensure dataset exists (dry_run gates the underlying create_dataset call)
         dataset = self._ensure_dataset(
diff --git a/src/rosetta-cli/tests/test_ragflow_client_dataset_name_resolution.py b/src/rosetta-cli/tests/test_ragflow_client_dataset_name_resolution.py
new file mode 100644
index 00000000..c1319418
--- /dev/null
+++ b/src/rosetta-cli/tests/test_ragflow_client_dataset_name_resolution.py
@@ -0,0 +1,93 @@
+from unittest.mock import Mock
+
+import pytest
+
+from rosetta_cli.ragflow_client import DocumentMetadata, RAGFlowClient
+
+
+def _make_client() -> RAGFlowClient:
+    client = object.__new__(RAGFlowClient)
+    client._client = None
+    client.dataset_default = "aia"
+    client._ensure_dataset = Mock(return_value=None)
+    return client
+
+
+def _make_metadata(**overrides) -> DocumentMetadata:
+    base = dict(
+        tags=["a", "b"],
+        domain="d",
+        release="r2",
+        content_hash="h" * 32,
+        ims_doc_id="ims-1",
+    )
+    base.update(overrides)
+    return DocumentMetadata(**base)
+
+
+def test_upload_document_uses_default_dataset_name_when_release_is_empty():
+    client = _make_client()
+
+    client.upload_document(
+        file_path=None,
+        metadata=_make_metadata(release=""),
+        dataset_name="aia",
+        dataset_template="aia-{release}",
+        content=b"hello",
+        dry_run=True,
+    )
+
+    resolved_name = client._ensure_dataset.call_args.args[0]
+
+    assert resolved_name == "aia"
+
+
+def test_upload_document_resolves_dataset_name_when_release_is_present():
+    client = _make_client()
+
+    client.upload_document(
+        file_path=None,
+        metadata=_make_metadata(),
+        dataset_name="aia",
+        dataset_template="aia-{release}",
+        content=b"hello",
+        dry_run=True,
+    )
+
+    resolved_name = client._ensure_dataset.call_args.args[0]
+
+    assert resolved_name == "aia-r2"
+
+
+def test_upload_document_uses_dataset_default_when_dataset_name_is_none():
+    client = _make_client()
+
+    client.upload_document(
+        file_path=None,
+        metadata=_make_metadata(release=""),
+        dataset_name=None,
+        dataset_template="aia-{release}",
+        content=b"hello",
+        dry_run=True,
+    )
+
+    resolved_name = client._ensure_dataset.call_args.args[0]
+
+    assert resolved_name == "aia"
+
+
+def test_upload_document_rejects_unresolved_release_placeholder():
+    client = _make_client()
+
+    with pytest.raises(
+        ValueError,
+        match="Unresolved release placeholder",
+    ):
+        client.upload_document(
+            file_path=None,
+            metadata=_make_metadata(release=""),
+            dataset_name="custom-{release}",
+            dataset_template="aia-{release}",
+            content=b"hello",
+            dry_run=True,
+        )
diff --git a/src/rosetta-cli/tests/test_ragflow_client_meta_fields_v25.py b/src/rosetta-cli/tests/test_ragflow_client_meta_fields_v25.py
index 71b4ba26..06ff61ac 100644
--- a/src/rosetta-cli/tests/test_ragflow_client_meta_fields_v25.py
+++ b/src/rosetta-cli/tests/test_ragflow_client_meta_fields_v25.py
@@ -49,6 +49,7 @@ def _make_client(dataset: _FakeDataset, list_documents_returns=()):
     client = object.__new__(RAGFlowClient)
     client._client = None
     client.page_size = 1000
+    client.dataset_default = "aia"
     client._doc_index_by_dataset = {}
     client._ensure_dataset = lambda *_a, **_k: dataset
     client.list_documents = lambda *_a, **_k: list(list_documents_returns)
diff --git a/src/rosetta-cli/tests/test_ragflow_client_upload_exception_handling.py b/src/rosetta-cli/tests/test_ragflow_client_upload_exception_handling.py
index 4f4ceede..2b8c9fc9 100644
--- a/src/rosetta-cli/tests/test_ragflow_client_upload_exception_handling.py
+++ b/src/rosetta-cli/tests/test_ragflow_client_upload_exception_handling.py
@@ -29,6 +29,7 @@ def _make_client():
     client = object.__new__(RAGFlowClient)
     client._client = None
     client.page_size = 1000
+    client.dataset_default = "aia"
     client._doc_index_by_dataset = {}
     return client