fix!: default vector_stores_config to VectorStoresConfig() to prevent file_search crash (backport #5949) (#5951)

When `vector_stores_config` was not set in the responses provider
config, `file_search` tool calls crashed with "'NoneType' object has no
attribute 'file_search_params'" because the config defaulted to `None`.
Since `VectorStoresConfig` has sensible defaults for all fields, default
it to `VectorStoresConfig()` in the config. Coerce `None` to
`VectorStoresConfig()` in the `ToolExecutor` constructor so that tests
do not have to explicitly pass it.

Minor breaking change: users that previously set an explicit null value
for vector stores config (`vector_stores_config: null`) will now
encounter validation errors. This is unlikely to affect anyone.

Signed-off-by: Eoin Fennessy <efenness@redhat.com>



---

<a href="https://app.devin.ai/review/ogx-ai/ogx/pull/5949"
target="_blank">
  <picture>
<source media="(prefers-color-scheme: dark)"
srcset="https://static.devin.ai/assets/gh-open-in-devin-review-dark.svg?v=1">
<img
src="https://static.devin.ai/assets/gh-open-in-devin-review-light.svg?v=1"
alt="Open in Devin Review">
  </picture>
</a>
<hr>This is an automatic backport of pull request #5949 done by
[Mergify](https://mergify.com).
<!-- devin-review-badge-begin -->

---

<a href="https://app.devin.ai/review/ogx-ai/ogx/pull/5951"
target="_blank">
  <picture>
<source media="(prefers-color-scheme: dark)"
srcset="https://static.devin.ai/assets/gh-open-in-devin-review-dark.svg?v=1">
<img
src="https://static.devin.ai/assets/gh-open-in-devin-review-light.svg?v=1"
alt="Open in Devin Review">
  </picture>
</a>
<!-- devin-review-badge-end -->

Signed-off-by: Eoin Fennessy <efenness@redhat.com>
Co-authored-by: Eoin Fennessy <efenness@redhat.com>
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

Author: 3 people

docs/docs/providers/responses/inline_builtin.mdx

@@ -20,7 +20,7 @@ Meta's reference implementation of an agent system that can use tools, access ve
 | `persistence.responses.backend` | `str` | No |  | Name of backend from storage.backends |
 | `persistence.responses.max_write_queue_size` | `int` | No | 10000 | Max queued writes for inference store |
 | `persistence.responses.num_writers` | `int` | No | 4 | Number of concurrent background writers |
-| `vector_stores_config` | `VectorStoresConfig \| None` | No |  | Configuration for vector store prompt templates and behavior |
+| `vector_stores_config` | `VectorStoresConfig` | No | default_provider_id=None default_embedding_model=None default_reranker_model=None rewrite_query_params=None file_search_params=FileSearchParams(header_template='file_search tool found {num_chunks} chunks:\nBEGIN of file_search tool results.\n', footer_template='END of file_search tool results.\n') context_prompt_params=ContextPromptParams(chunk_annotation_template='Result {index}\nContent: {chunk.content}\nMetadata: {metadata}\n', context_template='The above results were retrieved to help answer the user\'s query: "{query}". Use them as supporting information only in answering this query. {annotation_instruction}\n') annotation_prompt_params=AnnotationPromptParams(enable_annotations=True, annotation_instruction_template="Cite sources immediately at the end of sentences before punctuation, using `<|file-id|>` format like 'This is a fact <|file-Cn3MSNn72ENTiiq11Qda4A|>.'. Do not add extra punctuation. Use only the file IDs provided, do not invent new ones.", chunk_annotation_template='[{index}] {metadata_text} cite as <|{file_id}|>\n{chunk_text}\n') file_ingestion_params=FileIngestionParams(default_chunk_size_tokens=512, default_chunk_overlap_tokens=128) chunk_retrieval_params=ChunkRetrievalParams(chunk_multiplier=5, max_tokens_in_context=4000, default_reranker_strategy='rrf', rrf_impact_factor=60.0, weighted_search_alpha=0.5, default_search_mode='vector') file_batch_params=FileBatchParams(max_concurrent_files_per_batch=3, file_batch_chunk_size=10, cleanup_interval_seconds=86400) contextual_retrieval_params=ContextualRetrievalParams(model=None, default_timeout_seconds=120, default_max_concurrency=3, max_document_tokens=100000) | Configuration for vector store prompt templates and behavior |
 | `vector_stores_config.default_provider_id` | `str \| None` | No |  | ID of the vector_io provider to use as default when multiple providers are available and none is specified. |
 | `vector_stores_config.default_embedding_model` | `QualifiedModel \| None` | No |  | Default embedding model configuration for vector stores. |
 | `vector_stores_config.default_embedding_model.provider_id` | `str` | No |  |  |

src/ogx/providers/inline/responses/builtin/config.py

@@ -67,8 +67,8 @@ class BuiltinResponsesImplConfig(BaseModel):
 
     persistence: ResponsesPersistenceConfig
 
-    vector_stores_config: VectorStoresConfig | None = Field(
-        default=None,
+    vector_stores_config: VectorStoresConfig = Field(
+        default_factory=VectorStoresConfig,
         description="Configuration for vector store prompt templates and behavior",
     )
 

src/ogx/providers/inline/responses/builtin/responses/openai_responses.py

@@ -15,6 +15,7 @@
 from pydantic import TypeAdapter
 
 from ogx.core.conversations.validation import CONVERSATION_ID_PATTERN
+from ogx.core.datatypes import VectorStoresConfig
 from ogx.core.task import (
     RequestContext,
     activate_request_context,
@@ -117,7 +118,7 @@ def __init__(
         prompts_api: Prompts,
         files_api: Files,
         connectors_api: Connectors,
-        vector_stores_config=None,
+        vector_stores_config: VectorStoresConfig | None = None,
         compaction_config=None,
     ):
         self.inference_api = inference_api

src/ogx/providers/inline/responses/builtin/responses/tool_executor.py

@@ -56,13 +56,13 @@ def __init__(
         tool_groups_api: ToolGroups,
         tool_runtime_api: ToolRuntime,
         vector_io_api: VectorIO,
-        vector_stores_config=None,
+        vector_stores_config: VectorStoresConfig | None = None,
         mcp_session_manager=None,
     ):
         self.tool_groups_api = tool_groups_api
         self.tool_runtime_api = tool_runtime_api
         self.vector_io_api = vector_io_api
-        self.vector_stores_config = vector_stores_config
+        self.vector_stores_config = vector_stores_config or VectorStoresConfig()
         # Optional MCPSessionManager for session reuse within a request (fix for #4452)
         self.mcp_session_manager = mcp_session_manager
 
@@ -136,14 +136,7 @@ async def _execute_file_search_via_vector_store(
         # Create search tasks for all vector stores
         async def search_single_store(vector_store_id):
             try:
-                # Use default_search_mode from config if available
-                search_mode = "vector"
-                if (
-                    self.vector_stores_config
-                    and hasattr(self.vector_stores_config, "chunk_retrieval_params")
-                    and hasattr(self.vector_stores_config.chunk_retrieval_params, "default_search_mode")
-                ):
-                    search_mode = self.vector_stores_config.chunk_retrieval_params.default_search_mode
+                search_mode = self.vector_stores_config.chunk_retrieval_params.default_search_mode
 
                 search_response = await self.vector_io_api.openai_search_vector_store(
                     vector_store_id=vector_store_id,
@@ -171,12 +164,7 @@ async def search_single_store(vector_store_id):
 
         # Get templates from vector stores config, fallback to constants
 
-        # Check if annotations are enabled
-        enable_annotations = (
-            self.vector_stores_config
-            and self.vector_stores_config.annotation_prompt_params
-            and self.vector_stores_config.annotation_prompt_params.enable_annotations
-        )
+        enable_annotations = self.vector_stores_config.annotation_prompt_params.enable_annotations
 
         # Get templates
         header_template = self.vector_stores_config.file_search_params.header_template

tests/unit/providers/responses/builtin/test_openai_responses_tools.py

@@ -558,6 +558,42 @@ async def test_file_search_results_include_chunk_metadata_attributes(mock_vector
     ]
 
 
+async def test_file_search_works_without_explicit_vector_stores_config(mock_vector_io_api):
+    """Test that file_search works when vector_stores_config is not passed to ToolExecutor."""
+    query = "What is machine learning?"
+    vector_store_id = "test_vector_store"
+
+    mock_vector_io_api.openai_search_vector_store.return_value = VectorStoreSearchResponsePage(
+        search_query=[query],
+        has_more=False,
+        data=[
+            VectorStoreSearchResponse(
+                file_id="doc-123",
+                filename="ml-intro.md",
+                content=[VectorStoreContent(type="text", text="Machine learning is a subset of AI")],
+                score=0.95,
+                attributes={},
+            ),
+        ],
+    )
+
+    tool_executor = ToolExecutor(
+        tool_groups_api=None,  # type: ignore
+        tool_runtime_api=None,  # type: ignore
+        vector_io_api=mock_vector_io_api,
+        mcp_session_manager=None,
+    )
+
+    file_search_tool = OpenAIResponseInputToolFileSearch(vector_store_ids=[vector_store_id])
+    result = await tool_executor._execute_file_search_via_vector_store(
+        query=query,
+        response_file_search_tool=file_search_tool,
+    )
+
+    mock_vector_io_api.openai_search_vector_store.assert_called_once()
+    assert result.content is not None
+
+
 async def test_tool_call_arguments_arrive_in_subsequent_delta(openai_responses_impl, mock_inference_api):
     """Test that tool call arguments are correctly accumulated when the model streams
     arguments=None in the first delta and actual arguments in a subsequent delta.