diff --git a/api/controllers/console/knowledge_fs/resources.py b/api/controllers/console/knowledge_fs/resources.py
index 83362e3a39f..67a3b022dc3 100644
--- a/api/controllers/console/knowledge_fs/resources.py
+++ b/api/controllers/console/knowledge_fs/resources.py
@@ -240,6 +240,7 @@ from services.knowledge_fs.product_dto import (
     KnowledgeFSStreamCapabilityResponse,
     KnowledgeFSTraceEntriesQuery,
     KnowledgeFSTraceEntryListResponse,
+    KnowledgeFSTraceListQuery,
     KnowledgeFSTraceListResponse,
     KnowledgeFSUploadPartPresignPayload,
     KnowledgeFSUploadSessionAbortPayload,
@@ -3351,7 +3352,7 @@ class KnowledgeFSSpaceResearchTaskPartialsApi(Resource):
 
 @console_ns.route("/knowledge-fs/spaces/<string:control_space_id>/traces")
 class KnowledgeFSSpaceTracesApi(Resource):
-    @console_ns.doc(params=query_params_from_model(KnowledgeFSCursorQuery))
+    @console_ns.doc(params=query_params_from_model(KnowledgeFSTraceListQuery))
     @console_ns.response(
         HTTPStatus.OK,
         "KnowledgeFS traces",
@@ -3363,12 +3364,13 @@ class KnowledgeFSSpaceTracesApi(Resource):
     @_knowledge_fs_errors
     def get(self, control_space_id: str):
         actor_id, tenant_id = _actor()
-        query = KnowledgeFSCursorQuery.model_validate(request.args.to_dict())
+        query = KnowledgeFSTraceListQuery.model_validate(request.args.to_dict())
         result = _console_services().facade.list_traces(
             tenant_id=tenant_id,
             account_id=actor_id,
             control_space_id=control_space_id,
             cursor=query.cursor,
+            source=query.source,
         )
         return dump_response(KnowledgeFSTraceListResponse, result)
 
diff --git a/api/controllers/service_api/knowledge_fs/resources.py b/api/controllers/service_api/knowledge_fs/resources.py
index a7ea7bd76c4..b4351ab8d89 100644
--- a/api/controllers/service_api/knowledge_fs/resources.py
+++ b/api/controllers/service_api/knowledge_fs/resources.py
@@ -84,6 +84,7 @@ from services.knowledge_fs.product_dto import (
     KnowledgeFSSourceUpdatePayload,
     KnowledgeFSTraceEntriesQuery,
     KnowledgeFSTraceEntryListResponse,
+    KnowledgeFSTraceListQuery,
     KnowledgeFSTraceListResponse,
 )
 from services.knowledge_fs.product_operations import product_operation_action
@@ -955,7 +956,7 @@ class KnowledgeFSServiceResearchTaskPartialsApi(Resource):
 
 @service_api_ns.route("/knowledge-fs/spaces/<string:control_space_id>/traces")
 class KnowledgeFSServiceTracesApi(Resource):
-    @service_api_ns.doc(params=query_params_from_model(KnowledgeFSCursorQuery))
+    @service_api_ns.doc(params=query_params_from_model(KnowledgeFSTraceListQuery))
     @service_api_ns.response(
         HTTPStatus.OK,
         "KnowledgeFS traces",
@@ -965,11 +966,11 @@ class KnowledgeFSServiceTracesApi(Resource):
     def get(self, control_space_id: str):
         runtime = _runtime()
         profile = _profile(runtime, operation_id="listTraces", control_space_id=control_space_id)
-        query = KnowledgeFSCursorQuery.model_validate(request.args.to_dict())
+        query = KnowledgeFSTraceListQuery.model_validate(request.args.to_dict())
         raw = runtime.facade.execute_service(
             profile=profile,
             operation_id="listTraces",
-            query=(("cursor", query.cursor),) if query.cursor else (),
+            query=tuple((name, value) for name, value in (("cursor", query.cursor), ("source", query.source)) if value),
         )
         return dump_response(KnowledgeFSTraceListResponse, raw)
 
diff --git a/api/core/workflow/nodes/knowledge_retrieval_v2/knowledge_retrieval_v2_node.py b/api/core/workflow/nodes/knowledge_retrieval_v2/knowledge_retrieval_v2_node.py
index a81c4dd65ea..6cd5b9e8aa1 100644
--- a/api/core/workflow/nodes/knowledge_retrieval_v2/knowledge_retrieval_v2_node.py
+++ b/api/core/workflow/nodes/knowledge_retrieval_v2/knowledge_retrieval_v2_node.py
@@ -6,7 +6,7 @@ import logging
 import math
 import time
 import uuid
-from collections.abc import Iterator, Mapping, Sequence
+from collections.abc import Generator, Mapping, Sequence
 from concurrent.futures import ThreadPoolExecutor
 from contextlib import contextmanager
 from dataclasses import dataclass
@@ -409,7 +409,7 @@ class KnowledgeRetrievalV2Node(Node[KnowledgeRetrievalV2NodeData]):
 
     @staticmethod
     @contextmanager
-    def _map_space_errors(control_space_id: str, *, action: str) -> Iterator[None]:
+    def _map_space_errors(control_space_id: str, *, action: str) -> Generator[None]:
         try:
             yield
         except KnowledgeFSAppChannelDisabledError as exc:
@@ -807,7 +807,9 @@ class KnowledgeRetrievalV2Node(Node[KnowledgeRetrievalV2NodeData]):
                     control_space_id=control_space_id,
                     query=query,
                     mode=response.mode,
-                    retrieval_trace_id=response.trace_id,
+                    # Prefer the recorded history trace so the capture attaches to it instead of
+                    # creating a second record; older gateways only return the transport trace.
+                    retrieval_trace_id=response.answer_trace_id or response.trace_id,
                 )
             except Exception:
                 # The helper owns broker failures in production. Keep a second boundary here so a
diff --git a/api/knowledge-fs-contract.lock.json b/api/knowledge-fs-contract.lock.json
index 86e8b139c14..6b1aa89559e 100644
--- a/api/knowledge-fs-contract.lock.json
+++ b/api/knowledge-fs-contract.lock.json
@@ -1,7 +1,7 @@
 {
   "schemaVersion": 5,
-  "subtreeTree": "219d157af4b240de61ce77c7c3473bdca699addd",
-  "openapiSha256": "edeae0b8fd3bd6dc9512b396da508bb99862b12ad0faeebc130996c108496d85",
+  "subtreeTree": "dd565881a8b5a5c113292abeaef493579eccf97c",
+  "openapiSha256": "20e7f698679f6c9b51aba2cc386e49eee1fbc76d05b1c0c35609963c5fcc105a",
   "capabilityV2AuthManifestSha256": "e322a2fa779d1f40b95c54c1021cffecaec77abbd7b34899573dcdf4ff353109",
   "capabilityV2AuthTestVectorSha256": "ae0de37b1ff05c40f905cf17a7b410d8971acacf64db07d5ee3d6fecfa559ce3",
   "productOperationManifestSha256": "e4611c61c69478f20c1e1e482c44d8c48e48adfc67f1e8941ce36225e10b32c0",
diff --git a/api/openapi/markdown/console-openapi.md b/api/openapi/markdown/console-openapi.md
index a08380d1e26..60315c22924 100644
--- a/api/openapi/markdown/console-openapi.md
+++ b/api/openapi/markdown/console-openapi.md
@@ -9053,6 +9053,7 @@ Claim a workspace-staged upload. Multipart file bodies remain accepted as a lega
 | Name | Located in | Description | Required | Schema |
 | ---- | ---------- | ----------- | -------- | ------ |
 | cursor | query |  | No | string |
+| source | query |  | No | string, <br>**Available values:** "agent", "mcp", "retrieval_test", "service_api", "workflow" |
 | control_space_id | path |  | Yes | string |
 
 #### Responses
@@ -21056,6 +21057,7 @@ Input field definition for snippet parameters.
 | mode | string, <br>**Available values:** "auto", "deep", "fast", "research" | *Enum:* `"auto"`, `"deep"`, `"fast"`, `"research"` | Yes |
 | query | string |  | Yes |
 | query_images | [ [KnowledgeFSQueryImageResponse](#knowledgefsqueryimageresponse) ] |  | No |
+| source | [KnowledgeFSTraceSource](#knowledgefstracesource) |  | No |
 | steps | [ [KnowledgeFSAnswerTraceStepResponse](#knowledgefsanswertracestepresponse) ] |  | Yes |
 
 #### KnowledgeFSAnswerTraceStepResponse
@@ -21344,10 +21346,20 @@ Input field definition for snippet parameters.
 | ---- | ---- | ----------- | -------- |
 | KnowledgeFSControlSpaceVisibility | string |  |  |
 
+#### KnowledgeFSCrawlImportPagePayload
+
+| Name | Type | Description | Required |
+| ---- | ---- | ----------- | -------- |
+| content | string |  | Yes |
+| description | string |  | No |
+| sourceUrl | string |  | Yes |
+| title | string |  | No |
+
 #### KnowledgeFSCrawlImportPayload
 
 | Name | Type | Description | Required |
 | ---- | ---- | ----------- | -------- |
+| pages | [ [KnowledgeFSCrawlImportPagePayload](#knowledgefscrawlimportpagepayload) ] |  | No |
 | sourceUrls | [ string ] |  | Yes |
 
 #### KnowledgeFSCrawlPreviewPageListQuery
@@ -21918,6 +21930,7 @@ Input field definition for snippet parameters.
 | Name | Type | Description | Required |
 | ---- | ---- | ----------- | -------- |
 | description | string |  | No |
+| page_id | string |  | No |
 | source_url | string |  | Yes |
 | title | string |  | No |
 
@@ -21956,6 +21969,7 @@ Input field definition for snippet parameters.
 | Name | Type | Description | Required |
 | ---- | ---- | ----------- | -------- |
 | canonical_url | string |  | No |
+| pageId | string |  | No |
 | source_url | string |  | Yes |
 | title | string |  | No |
 
@@ -21972,6 +21986,7 @@ Input field definition for snippet parameters.
 | parameters | object |  | No |
 | pluginId | string |  | No |
 | previewConfigurationFingerprint | string |  | No |
+| previewJobId | string |  | No |
 | provider | string |  | Yes |
 | providerDisplayName | string |  | No |
 | root_url | string |  | Yes |
@@ -22390,9 +22405,17 @@ Input field definition for snippet parameters.
 
 #### KnowledgeFSProductPermission
 
+Product-level capability of a caller on one KnowledgeFS space.
+
+These values are what the console returns as ``permission_keys`` so the web can gate
+its UI. Enterprise RBAC is not evaluated in this vocabulary: KnowledgeFS spaces reuse
+the legacy knowledge base (``dataset_*``) permission points, and each capability is
+granted when the caller holds the dataset permission it maps to (see
+:data:`RBAC_PERMISSION_BY_PRODUCT_PERMISSION`).
+
 | Name | Type | Description | Required |
 | ---- | ---- | ----------- | -------- |
-| KnowledgeFSProductPermission | string |  |  |
+| KnowledgeFSProductPermission | string | Product-level capability of a caller on one KnowledgeFS space.  These values are what the console returns as ``permission_keys`` so the web can gate its UI. Enterprise RBAC is not evaluated in this vocabulary: KnowledgeFS spaces reuse the legacy knowledge base (``dataset_*``) permission points, and each capability is granted when the caller holds the dataset permission it maps to (see :data:`RBAC_PERMISSION_BY_PRODUCT_PERMISSION`). |  |
 
 #### KnowledgeFSProductRerankProfile
 
@@ -23476,6 +23499,7 @@ the file name and a short-lived signed preview URL for files the caller still ow
 | query_images | [ [KnowledgeFSQueryImageResponse](#knowledgefsqueryimageresponse) ] |  | No |
 | result_count | integer |  | Yes |
 | scores | [KnowledgeFSTraceScoresResponse](#knowledgefstracescoresresponse) |  | Yes |
+| source | [KnowledgeFSTraceSource](#knowledgefstracesource) |  | No |
 | stages | [ [KnowledgeFSTraceStageResponse](#knowledgefstracestageresponse) ] |  | Yes |
 
 #### KnowledgeFSTraceScoresResponse
@@ -23486,6 +23510,12 @@ the file name and a short-lived signed preview URL for files the caller still ow
 | rerank | number |  | No |
 | retrieval | number |  | No |
 
+#### KnowledgeFSTraceSource
+
+| Name | Type | Description | Required |
+| ---- | ---- | ----------- | -------- |
+| KnowledgeFSTraceSource | string |  |  |
+
 #### KnowledgeFSTraceStageResponse
 
 | Name | Type | Description | Required |
diff --git a/api/openapi/markdown/service-openapi.md b/api/openapi/markdown/service-openapi.md
index 56703a961bc..1697ff31d91 100644
--- a/api/openapi/markdown/service-openapi.md
+++ b/api/openapi/markdown/service-openapi.md
@@ -663,6 +663,7 @@ Deprecated legacy alias for updating an existing document by providing text cont
 | Name | Located in | Description | Required | Schema |
 | ---- | ---------- | ----------- | -------- | ------ |
 | cursor | query |  | No | string |
+| source | query |  | No | string, <br>**Available values:** "agent", "mcp", "retrieval_test", "service_api", "workflow" |
 | control_space_id | path |  | Yes | string |
 
 #### Responses
@@ -4274,6 +4275,7 @@ Model class for i18n object.
 | mode | string, <br>**Available values:** "auto", "deep", "fast", "research" | *Enum:* `"auto"`, `"deep"`, `"fast"`, `"research"` | Yes |
 | query | string |  | Yes |
 | query_images | [ [KnowledgeFSQueryImageResponse](#knowledgefsqueryimageresponse) ] |  | No |
+| source | [KnowledgeFSTraceSource](#knowledgefstracesource) |  | No |
 | steps | [ [KnowledgeFSAnswerTraceStepResponse](#knowledgefsanswertracestepresponse) ] |  | Yes |
 
 #### KnowledgeFSAnswerTraceStepResponse
@@ -5223,6 +5225,7 @@ the file name and a short-lived signed preview URL for files the caller still ow
 | query_images | [ [KnowledgeFSQueryImageResponse](#knowledgefsqueryimageresponse) ] |  | No |
 | result_count | integer |  | Yes |
 | scores | [KnowledgeFSTraceScoresResponse](#knowledgefstracescoresresponse) |  | Yes |
+| source | [KnowledgeFSTraceSource](#knowledgefstracesource) |  | No |
 | stages | [ [KnowledgeFSTraceStageResponse](#knowledgefstracestageresponse) ] |  | Yes |
 
 #### KnowledgeFSTraceScoresResponse
@@ -5233,6 +5236,12 @@ the file name and a short-lived signed preview URL for files the caller still ow
 | rerank | number |  | No |
 | retrieval | number |  | No |
 
+#### KnowledgeFSTraceSource
+
+| Name | Type | Description | Required |
+| ---- | ---- | ----------- | -------- |
+| KnowledgeFSTraceSource | string |  |  |
+
 #### KnowledgeFSTraceStageResponse
 
 | Name | Type | Description | Required |
diff --git a/api/services/knowledge_fs/data_facade.py b/api/services/knowledge_fs/data_facade.py
index a043287e6da..e0ab9adb217 100644
--- a/api/services/knowledge_fs/data_facade.py
+++ b/api/services/knowledge_fs/data_facade.py
@@ -2079,14 +2079,20 @@ class KnowledgeFSDataFacade:
         return tasks
 
     def list_traces(
-        self, *, tenant_id: str, account_id: str, control_space_id: str, cursor: str | None = None
+        self,
+        *,
+        tenant_id: str,
+        account_id: str,
+        control_space_id: str,
+        cursor: str | None = None,
+        source: str | None = None,
     ) -> KnowledgeFSTraceListResponse:
         raw = self._interactive(
             tenant_id=tenant_id,
             account_id=account_id,
             control_space_id=control_space_id,
             operation_id="listTraces",
-            query=(("cursor", cursor),) if cursor else (),
+            query=_trace_list_query(cursor=cursor, source=source),
         )
         traces = KnowledgeFSTraceListResponse.model_validate(raw)
         self._attach_query_image_previews(
@@ -2738,3 +2744,12 @@ def _assert_sse_bff_ready(operation_id: str) -> None:
 
 
 __all__ = ["KnowledgeFSDataFacade"]
+
+
+def _trace_list_query(*, cursor: str | None, source: str | None) -> tuple[tuple[str, str], ...]:
+    query: list[tuple[str, str]] = []
+    if cursor:
+        query.append(("cursor", cursor))
+    if source:
+        query.append(("source", source))
+    return tuple(query)
diff --git a/api/services/knowledge_fs/product_dto.py b/api/services/knowledge_fs/product_dto.py
index a2fc86dc74d..b18e6912799 100644
--- a/api/services/knowledge_fs/product_dto.py
+++ b/api/services/knowledge_fs/product_dto.py
@@ -498,6 +498,15 @@ class KnowledgeFSCursorQuery(BaseModel):
     model_config = ConfigDict(extra="forbid")
 
 
+# Who produced a retrieval trace: the console retrieval test, a KnowledgeFS workflow node, the
+# service API, an agent, or an MCP client. Mirrors KnowledgeFS `AnswerTraceSourceSchema`.
+type KnowledgeFSTraceSource = Literal["retrieval_test", "workflow", "service_api", "agent", "mcp"]
+
+
+class KnowledgeFSTraceListQuery(KnowledgeFSCursorQuery):
+    source: KnowledgeFSTraceSource | None = Field(default=None)
+
+
 type KnowledgeFSConsistencyClass = Literal[
     "path-consistent",
     "snapshot-consistent",
@@ -3133,6 +3142,9 @@ class KnowledgeFSRetrievalTestMetricsResponse(ResponseModel):
 
 
 class KnowledgeFSRetrievalTestResponse(ResponseModel):
+    # AnswerTrace recorded for this run in the space's retrieval history; a failed-retrieval capture
+    # attaches to it so one record stays per retrieval.
+    answer_trace_id: str | None = Field(default=None, validation_alias=AliasChoices("answer_trace_id", "answerTraceId"))
     items: list[KnowledgeFSRetrievalTestItemResponse] = Field(max_length=100)
     metrics: KnowledgeFSRetrievalTestMetricsResponse
     mode: Literal["deep", "fast", "research"]
@@ -3334,6 +3346,7 @@ class KnowledgeFSAnswerTraceResponse(ResponseModel):
         validation_alias=AliasChoices("query_images", "queryImages"),
         exclude_if=lambda value: not value,
     )
+    source: KnowledgeFSTraceSource = "retrieval_test"
     steps: list[KnowledgeFSAnswerTraceStepResponse]
 
 
@@ -3423,6 +3436,7 @@ class KnowledgeFSTraceResponse(ResponseModel):
     )
     result_count: int = Field(ge=0, validation_alias=AliasChoices("result_count", "resultCount"))
     scores: KnowledgeFSTraceScoresResponse
+    source: KnowledgeFSTraceSource = "retrieval_test"
     stages: list[KnowledgeFSTraceStageResponse]
 
 
diff --git a/api/tests/unit_tests/core/workflow/nodes/knowledge_retrieval_v2/test_knowledge_retrieval_v2_node.py b/api/tests/unit_tests/core/workflow/nodes/knowledge_retrieval_v2/test_knowledge_retrieval_v2_node.py
index a4d8e5f9d70..1c9b15feec0 100644
--- a/api/tests/unit_tests/core/workflow/nodes/knowledge_retrieval_v2/test_knowledge_retrieval_v2_node.py
+++ b/api/tests/unit_tests/core/workflow/nodes/knowledge_retrieval_v2/test_knowledge_retrieval_v2_node.py
@@ -1328,3 +1328,22 @@ def test_multi_space_calls_are_concurrent_but_never_exceed_the_bound() -> None:
 
     assert result.status == WorkflowNodeExecutionStatus.SUCCEEDED
     assert service.max_active == 4
+
+
+def test_quality_capture_prefers_the_recorded_history_trace_over_the_transport_trace(
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    """Newer gateways return the AnswerTrace id so the capture attaches to the same history record."""
+    dispatched: list[dict[str, object]] = []
+    monkeypatch.setattr(
+        node_module,
+        "enqueue_workflow_failed_retrieval_capture",
+        lambda **kwargs: dispatched.append(kwargs),
+    )
+    recorded = _empty_response(mode="fast", space="a").model_copy(
+        update={"answer_trace_id": "018f0d60-7a49-7cc2-9c1b-5b36f18f2c60"}
+    )
+    result = _node(service=RecordingCapabilityService({"space-a": recorded}), spaces=["space-a"])._run()
+
+    assert result.status == WorkflowNodeExecutionStatus.SUCCEEDED
+    assert [call["retrieval_trace_id"] for call in dispatched] == ["018f0d60-7a49-7cc2-9c1b-5b36f18f2c60"]
diff --git a/api/tests/unit_tests/services/test_knowledge_fs_product_dto.py b/api/tests/unit_tests/services/test_knowledge_fs_product_dto.py
index d870aa9628a..cd56d6e31a3 100644
--- a/api/tests/unit_tests/services/test_knowledge_fs_product_dto.py
+++ b/api/tests/unit_tests/services/test_knowledge_fs_product_dto.py
@@ -62,6 +62,7 @@ from services.knowledge_fs.product_dto import (
     KnowledgeFSSpaceCreatePayload,
     KnowledgeFSSpaceListItemResponse,
     KnowledgeFSStatResponse,
+    KnowledgeFSTraceListQuery,
     KnowledgeFSTraceResponse,
     KnowledgeFSTreeResponse,
     KnowledgeFSUploadPartPresignPayload,
@@ -1641,3 +1642,38 @@ def test_overview_dtos_accept_the_knowledge_fs_wire_shape() -> None:
     assert KnowledgeFSOverviewWindowQuery().window == "24h"
     with pytest.raises(ValidationError):
         KnowledgeFSOverviewWindowQuery.model_validate({"window": "1h"})
+
+
+def test_trace_responses_carry_the_caller_source_and_default_to_retrieval_tests() -> None:
+    """Workflow-node retrievals share the history with console retrieval tests, distinguished by source."""
+    base = {
+        "completed": True,
+        "created_at": "2026-09-04T10:00:00.000Z",
+        "id": "trace-1",
+        "mode": "fast",
+        "profile": {},
+        "query": "why",
+        "result_count": 2,
+        "scores": {},
+        "stages": [],
+    }
+
+    assert KnowledgeFSTraceResponse.model_validate(base).source == "retrieval_test"
+    assert KnowledgeFSTraceResponse.model_validate({**base, "source": "workflow"}).model_dump(mode="json")[
+        "source"
+    ] == ("workflow")
+    assert KnowledgeFSTraceListQuery.model_validate({"source": "workflow"}).source == "workflow"
+    assert KnowledgeFSTraceListQuery.model_validate({}).source is None
+    with pytest.raises(ValidationError):
+        KnowledgeFSTraceListQuery.model_validate({"source": "console"})
+
+
+def test_retrieval_test_response_exposes_the_recorded_history_trace_id() -> None:
+    base = {"items": [], "metrics": {"degradationFlags": [], "totalMs": 4}, "mode": "fast", "traceId": "trace-a"}
+    assert KnowledgeFSRetrievalTestResponse.model_validate(base).answer_trace_id is None
+    assert (
+        KnowledgeFSRetrievalTestResponse.model_validate(
+            {**base, "answerTraceId": "018f0d60-7a49-7cc2-9c1b-5b36f18f2c60"}
+        ).answer_trace_id
+        == "018f0d60-7a49-7cc2-9c1b-5b36f18f2c60"
+    )
diff --git a/knowledge-fs/apps/api/src/index.ts b/knowledge-fs/apps/api/src/index.ts
index 05e910953fd..9e05906e7d8 100644
--- a/knowledge-fs/apps/api/src/index.ts
+++ b/knowledge-fs/apps/api/src/index.ts
@@ -943,6 +943,9 @@ const researchTaskRuntime =
             })}\n`,
           );
         },
+        ...(databaseRepositories.gatewayOptions.knowledgeSpaceOverview
+          ? { overview: databaseRepositories.gatewayOptions.knowledgeSpaceOverview }
+          : {}),
         partials: databaseRepositories.researchTaskPartialResults,
         progress: databaseRepositories.researchTaskProgressEvents,
         ...(researchProjectionSnapshotResolver
diff --git a/knowledge-fs/apps/api/src/migrate.test.ts b/knowledge-fs/apps/api/src/migrate.test.ts
index 6ddc9610f0a..c36b6cbc5be 100644
--- a/knowledge-fs/apps/api/src/migrate.test.ts
+++ b/knowledge-fs/apps/api/src/migrate.test.ts
@@ -153,6 +153,8 @@ describe("runApiDatabaseMigrations", () => {
       "insert",
       "schema",
       "insert",
+      "schema",
+      "insert",
     ]);
     expect(migrationSql).toHaveLength(expectedPostgresMigrationIds.length);
     expect(migrationSql[2]).toContain("-- Migration id: 0003_projection_set_publications\n");
@@ -236,6 +238,7 @@ describe("runApiDatabaseMigrations", () => {
     );
     expect(migrationSql[50]).toContain("-- Migration id: 0051_failed_query_trace_outcome_index\n");
     expect(migrationSql[51]).toContain("-- Migration id: 0052_namespace_source_previews\n");
+    expect(migrationSql[52]).toContain("-- Migration id: 0053_answer_trace_source\n");
     expect(closed).toBe(true);
   });
 
diff --git a/knowledge-fs/apps/api/src/research-task-runtime-options.ts b/knowledge-fs/apps/api/src/research-task-runtime-options.ts
index ddb92e439d8..6b1cbb05d4b 100644
--- a/knowledge-fs/apps/api/src/research-task-runtime-options.ts
+++ b/knowledge-fs/apps/api/src/research-task-runtime-options.ts
@@ -7,6 +7,7 @@ import {
   type KnowledgeGatewayOptions,
   type KnowledgeSpaceAccessService,
   type KnowledgeSpaceManifestRepository,
+  type KnowledgeSpaceOverviewRepository,
   type ModelInputModalityResolver,
   type PublishedProjectionReadSnapshotResolver,
   type QueryGenerator,
@@ -46,6 +47,8 @@ export interface CreateApiResearchTaskRuntimeOptions {
         readonly researchTaskJob?: { readonly id: string; readonly stage: string } | undefined;
       }) => void)
     | undefined;
+  /** Overview activity sink so durable Research tasks count as answered/failed queries. */
+  readonly overview?: Pick<KnowledgeSpaceOverviewRepository, "appendActivity"> | undefined;
   readonly partials: ResearchTaskPartialResultRepository;
   readonly progress: ResearchTaskProgressRepository;
   readonly projectionSnapshotResolver?: PublishedProjectionReadSnapshotResolver | undefined;
@@ -89,6 +92,7 @@ export function createApiResearchTaskRuntime({
   metrics,
   modelInputModalityResolver,
   onError,
+  overview,
   partials,
   progress,
   projectionSnapshotResolver,
@@ -120,6 +124,7 @@ export function createApiResearchTaskRuntime({
     ...(metrics ? { metrics } : {}),
     ...(modelInputModalityResolver ? { modelInputModalityResolver } : {}),
     ...(onError ? { onError } : {}),
+    ...(overview ? { overview } : {}),
     repository,
   });
   const runtime = createResearchTaskRuntime({
diff --git a/knowledge-fs/packages/adapters/src/migration-runner.test.ts b/knowledge-fs/packages/adapters/src/migration-runner.test.ts
index f389123db29..e5ff6c2db1a 100644
--- a/knowledge-fs/packages/adapters/src/migration-runner.test.ts
+++ b/knowledge-fs/packages/adapters/src/migration-runner.test.ts
@@ -60,6 +60,7 @@ const answerTraceQueryImagesMigrationId = "0049_answer_trace_query_images";
 const deletionJobCapabilityProvenanceMigrationId = "0050_deletion_job_capability_provenance";
 const failedQueryTraceOutcomeIndexMigrationId = "0051_failed_query_trace_outcome_index";
 const namespaceSourcePreviewsMigrationId = "0052_namespace_source_previews";
+const answerTraceSourceMigrationId = "0053_answer_trace_source";
 const migrationsAfterDurableDeletion = [
   versionedSpaceProfilesMigrationId,
   profilePublicationBindingsMigrationId,
@@ -96,6 +97,7 @@ const migrationsAfterDurableDeletion = [
   deletionJobCapabilityProvenanceMigrationId,
   failedQueryTraceOutcomeIndexMigrationId,
   namespaceSourcePreviewsMigrationId,
+  answerTraceSourceMigrationId,
 ] as const;
 const migrationsAfterTidbBaselineRepair = [
   spaceAccessControlMigrationId,
diff --git a/knowledge-fs/packages/api/src/answer-trace-recorder.ts b/knowledge-fs/packages/api/src/answer-trace-recorder.ts
index 3bcc4a7af27..a34b5b84682 100644
--- a/knowledge-fs/packages/api/src/answer-trace-recorder.ts
+++ b/knowledge-fs/packages/api/src/answer-trace-recorder.ts
@@ -1,6 +1,6 @@
 import { randomUUID } from "node:crypto";
 
-import { type AnswerTrace, AnswerTraceSchema } from "@knowledge/core";
+import { type AnswerTrace, AnswerTraceSchema, type AnswerTraceSource } from "@knowledge/core";
 import type { QueryImageMetadata } from "./query-images";
 
 import {
@@ -23,6 +23,8 @@ interface RecordAnswerTraceBaseInput {
   readonly mode: AnswerTrace["mode"];
   readonly query: string;
   readonly queryImages?: readonly QueryImageMetadata[] | undefined;
+  /** Caller that produced the trace; omitted means a console retrieval test. */
+  readonly source?: AnswerTraceSource | undefined;
   readonly steps: readonly RecordAnswerTraceStepInput[];
   readonly traceId?: string | undefined;
 }
diff --git a/knowledge-fs/packages/api/src/answer-trace-repository.test.ts b/knowledge-fs/packages/api/src/answer-trace-repository.test.ts
index f7fda991c51..0dd83950012 100644
--- a/knowledge-fs/packages/api/src/answer-trace-repository.test.ts
+++ b/knowledge-fs/packages/api/src/answer-trace-repository.test.ts
@@ -293,6 +293,7 @@ describe("AnswerTrace repositories", () => {
           true,
           trace.createdAt,
           null,
+          null,
         ],
         tableName: "answer_traces",
       }),
@@ -1219,3 +1220,58 @@ function hasWholeSpaceDeletionFence(
     )
   );
 }
+
+describe("AnswerTrace source column", () => {
+  it.each(["postgres", "tidb"] as const)("round-trips the trace source on %s", async (kind) => {
+    let storedTrace: DatabaseRow | undefined;
+    const inserts: DatabaseExecuteInput[] = [];
+    const executor = async (input: DatabaseExecuteInput): Promise<DatabaseExecuteResult> => {
+      if (input.tableName === "knowledge_spaces") {
+        return { rows: [{ id: input.params[0], tenant_id: "tenant-1" }], rowsAffected: 1 };
+      }
+      if (input.tableName === "answer_traces" && input.operation === "select") {
+        return { rows: storedTrace ? [storedTrace] : [], rowsAffected: storedTrace ? 1 : 0 };
+      }
+      if (input.tableName === "answer_traces" && input.operation === "insert") {
+        inserts.push(input);
+        storedTrace = {
+          access_channel: input.params[10],
+          capability_grant_id: input.params[3],
+          completed: input.params[11],
+          created_at: input.params[12],
+          evidence_bundle_id: input.params[4],
+          id: input.params[0],
+          knowledge_space_id: input.params[2],
+          mode: input.params[6],
+          permission_snapshot_id: input.params[8],
+          permission_snapshot_revision: input.params[9],
+          query: input.params[5],
+          query_images: input.params[13],
+          source: input.params[14],
+          subject_id: input.params[7],
+          tenant_id: input.params[1],
+        };
+        return { rows: [], rowsAffected: 1 };
+      }
+      return { rows: [], rowsAffected: input.operation === "insert" ? 1 : 0 };
+    };
+    const repository = createDatabaseAnswerTraceRepository({
+      database: createSchemaDatabaseAdapter({
+        executor,
+        kind,
+        transaction: async (callback) => callback({ execute: executor }),
+      }),
+    });
+    const trace = AnswerTraceSchema.parse({ ...databaseTrace(), source: "workflow" });
+
+    await repository.create(trace);
+
+    expect(inserts[0]?.sql).toContain(kind === "postgres" ? '"source"' : "`source`");
+    expect(inserts[0]?.params[14]).toBe("workflow");
+    expect((await repository.getById(trace.id))?.source).toBe("workflow");
+
+    // Rows written before the column existed read back without a source (= retrieval test).
+    storedTrace = { ...storedTrace, source: null };
+    expect((await repository.getById(trace.id))?.source).toBeUndefined();
+  });
+});
diff --git a/knowledge-fs/packages/api/src/answer-trace-repository.ts b/knowledge-fs/packages/api/src/answer-trace-repository.ts
index 0e31fb26c37..976319ab457 100644
--- a/knowledge-fs/packages/api/src/answer-trace-repository.ts
+++ b/knowledge-fs/packages/api/src/answer-trace-repository.ts
@@ -240,6 +240,7 @@ export function createDatabaseAnswerTraceRepository({
           "completed",
           "created_at",
           "query_images",
+          "source",
         ];
         const traceParams = [
           persistedTrace.id,
@@ -256,6 +257,7 @@ export function createDatabaseAnswerTraceRepository({
           answerTraceCompleted(persistedTrace),
           persistedTrace.createdAt,
           persistedTrace.queryImages?.length ? JSON.stringify(persistedTrace.queryImages) : null,
+          persistedTrace.source ?? null,
         ] satisfies readonly DatabaseQueryValue[];
         const admissionSpaceParameter =
           database.dialect === "postgres"
@@ -605,6 +607,7 @@ function mapAnswerTraceRows(traceRow: DatabaseRow, stepRows: readonly DatabaseRo
   );
   const accessChannel = optionalStringColumn(traceRow, "access_channel");
   const tenantId = optionalStringColumn(traceRow, "tenant_id");
+  const source = optionalStringColumn(traceRow, "source");
 
   return parseAnswerTraceProvenance({
     ...(capabilityGrantId === undefined ? {} : { capabilityGrantId }),
@@ -626,6 +629,7 @@ function mapAnswerTraceRows(traceRow: DatabaseRow, stepRows: readonly DatabaseRo
     ...(traceRow.query_images == null
       ? {}
       : { queryImages: jsonArrayColumn(traceRow, "query_images") }),
+    ...(source === undefined ? {} : { source }),
     ...(subjectId === undefined ? {} : { subjectId }),
     steps: stepRows.map((row) => ({
       endedAt: stringColumn(row, "ended_at"),
diff --git a/knowledge-fs/packages/api/src/answer-trace-source.test.ts b/knowledge-fs/packages/api/src/answer-trace-source.test.ts
new file mode 100644
index 00000000000..62c7b3e2abf
--- /dev/null
+++ b/knowledge-fs/packages/api/src/answer-trace-source.test.ts
@@ -0,0 +1,29 @@
+import { describe, expect, it } from "vitest";
+import { answerTraceSourceForCaller } from "./answer-trace-source";
+
+describe("answerTraceSourceForCaller", () => {
+  it("prefers the precise Capability v2 caller kind over the collapsed gateway kind", () => {
+    // The gateway maps workflow grants to the "agent" access channel; history must keep them apart.
+    expect(
+      answerTraceSourceForCaller({ callerKind: "agent", capabilityCallerKind: "workflow" }),
+    ).toBe("workflow");
+    expect(answerTraceSourceForCaller({ capabilityCallerKind: "agent" })).toBe("agent");
+    expect(answerTraceSourceForCaller({ capabilityCallerKind: "mcp" })).toBe("mcp");
+    expect(answerTraceSourceForCaller({ capabilityCallerKind: "service" })).toBe("service_api");
+    expect(answerTraceSourceForCaller({ capabilityCallerKind: "internal_worker" })).toBe(
+      "service_api",
+    );
+    expect(answerTraceSourceForCaller({ capabilityCallerKind: "interactive" })).toBe(
+      "retrieval_test",
+    );
+  });
+
+  it("falls back to the gateway caller kind for legacy sessions and API keys", () => {
+    expect(answerTraceSourceForCaller({ callerKind: "api_key" })).toBe("service_api");
+    expect(answerTraceSourceForCaller({ callerKind: "service_api" })).toBe("service_api");
+    expect(answerTraceSourceForCaller({ callerKind: "agent" })).toBe("agent");
+    expect(answerTraceSourceForCaller({ callerKind: "mcp" })).toBe("mcp");
+    expect(answerTraceSourceForCaller({ callerKind: "interactive" })).toBe("retrieval_test");
+    expect(answerTraceSourceForCaller({})).toBe("retrieval_test");
+  });
+});
diff --git a/knowledge-fs/packages/api/src/answer-trace-source.ts b/knowledge-fs/packages/api/src/answer-trace-source.ts
new file mode 100644
index 00000000000..c29d43f668d
--- /dev/null
+++ b/knowledge-fs/packages/api/src/answer-trace-source.ts
@@ -0,0 +1,45 @@
+import type { AnswerTraceSource } from "@knowledge/core";
+import type { DifyCapabilityV2SanitizedGrant } from "./dify-capability-v2-grant";
+import type { KnowledgeSpaceCallerKind } from "./knowledge-space-authorization";
+
+/**
+ * Derive the AnswerTrace source from the authenticated caller.
+ *
+ * Capability v2 grants carry the precise caller kind (the gateway collapses `workflow` into
+ * `agent` for access-channel purposes, which is too coarse for history); legacy sessions and
+ * API keys only expose the gateway caller kind. Anything unknown is a console retrieval test.
+ */
+export function answerTraceSourceForCaller({
+  callerKind,
+  capabilityCallerKind,
+}: {
+  readonly callerKind?: KnowledgeSpaceCallerKind | undefined;
+  readonly capabilityCallerKind?: DifyCapabilityV2SanitizedGrant["callerKind"] | undefined;
+}): AnswerTraceSource {
+  switch (capabilityCallerKind) {
+    case "workflow":
+      return "workflow";
+    case "agent":
+      return "agent";
+    case "mcp":
+      return "mcp";
+    case "service":
+    case "internal_worker":
+      return "service_api";
+    case "interactive":
+      return "retrieval_test";
+    default:
+      break;
+  }
+  switch (callerKind) {
+    case "agent":
+      return "agent";
+    case "mcp":
+      return "mcp";
+    case "api_key":
+    case "service_api":
+      return "service_api";
+    default:
+      return "retrieval_test";
+  }
+}
diff --git a/knowledge-fs/packages/api/src/gateway-sse-responses.ts b/knowledge-fs/packages/api/src/gateway-sse-responses.ts
index 35d1ec15c04..e90ae87cdbf 100644
--- a/knowledge-fs/packages/api/src/gateway-sse-responses.ts
+++ b/knowledge-fs/packages/api/src/gateway-sse-responses.ts
@@ -1,4 +1,5 @@
 import type {
+  AnswerTraceSource,
   AuthSubject,
   EvidenceBundle,
   KnowledgeSpaceEmbeddingProfile,
@@ -36,6 +37,8 @@ import {
 export type QueryGenerationMode = "deep" | "fast" | "research";
 
 export interface QueryGenerationInput {
+  /** Caller that produced this query; recorded on the AnswerTrace for history filtering. */
+  readonly source?: AnswerTraceSource | undefined;
   readonly capabilityGrantId?: string | undefined;
   readonly embeddingProfile?: KnowledgeSpaceEmbeddingProfile | undefined;
   readonly embeddingInputModalities?: readonly ModelInputModality[] | undefined;
@@ -472,6 +475,7 @@ async function recordAnswerTrace({
     mode: input.mode,
     query: input.query,
     ...(input.queryImageMetadata?.length ? { queryImages: input.queryImageMetadata } : {}),
+    ...(input.source ? { source: input.source } : {}),
     steps: [
       ...stageSteps,
       {
diff --git a/knowledge-fs/packages/api/src/gateway.test.ts b/knowledge-fs/packages/api/src/gateway.test.ts
index 9c15e4f9206..6da18a0337d 100644
--- a/knowledge-fs/packages/api/src/gateway.test.ts
+++ b/knowledge-fs/packages/api/src/gateway.test.ts
@@ -12872,7 +12872,7 @@ describe("createKnowledgeGateway", () => {
           trace.permissionSnapshot?.accessChannel,
           true,
           trace.createdAt,
-          null,
+          ...[null, trace.source ?? null],
         ],
         tableName: "answer_traces",
       }),
diff --git a/knowledge-fs/packages/api/src/index.ts b/knowledge-fs/packages/api/src/index.ts
index 10c4b70b4de..6530b2e37c4 100644
--- a/knowledge-fs/packages/api/src/index.ts
+++ b/knowledge-fs/packages/api/src/index.ts
@@ -1221,6 +1221,7 @@ export function createKnowledgeGateway({
     createResearchTaskJobStateMachine({
       generateId: generateResearchTaskJobId,
       jobs: adapter.jobs,
+      overview: overviewRepository,
       progress: createResearchTaskProgressPublisher({
         repository: researchTaskProgressEvents,
       }),
@@ -1984,6 +1985,7 @@ export function createKnowledgeGateway({
           }
         : {}),
     now: () => Date.parse(now()),
+    overview: overviewRepository,
     researchTaskJobs,
     researchTaskPartialResults,
     researchTaskProgressEvents,
diff --git a/knowledge-fs/packages/api/src/knowledge-space-overview.ts b/knowledge-fs/packages/api/src/knowledge-space-overview.ts
index 1a945044d18..6c3cccf8c3a 100644
--- a/knowledge-fs/packages/api/src/knowledge-space-overview.ts
+++ b/knowledge-fs/packages/api/src/knowledge-space-overview.ts
@@ -503,7 +503,11 @@ const SAFE_ACTIVITY_DETAIL_KEYS = new Set([
   "question",
   "questionTruncated",
   "reasonCode",
+  // Who issued a query (console retrieval test, workflow node, service API, agent, MCP).
+  "source",
   "statusCode",
+  // Distinguishes Research tasks from streamed queries in the query events.
+  "taskKind",
 ]);
 
 export function sanitizeKnowledgeSpaceActivityDetails(
diff --git a/knowledge-fs/packages/api/src/quality-control-database-repository.test.ts b/knowledge-fs/packages/api/src/quality-control-database-repository.test.ts
index 68fd648dedc..61bd00c184d 100644
--- a/knowledge-fs/packages/api/src/quality-control-database-repository.test.ts
+++ b/knowledge-fs/packages/api/src/quality-control-database-repository.test.ts
@@ -3008,3 +3008,144 @@ function permissionFenceRow(
     visibility: "all_members",
   };
 }
+
+describe("answer trace history sources", () => {
+  const WORKFLOW_TRACE_ID = "018f0d60-7a49-7cc2-9c1b-5b36f18f2c70";
+  const evidenceBundle = {
+    createdAt: NOW,
+    id: "018f0d60-7a49-7cc2-9c1b-5b36f18f2c71",
+    items: [
+      {
+        citations: [
+          { documentAssetId: "018f0d60-7a49-7cc2-9c1b-5b36f18f2c72", documentVersion: 1 },
+        ],
+        conflicts: [],
+        freshness: { status: "unknown" },
+        metadata: {},
+        nodeId: "018f0d60-7a49-7cc2-9c1b-5b36f18f2c73",
+        score: 0.8,
+        scores: { final: 0.8, retrieval: 0.7 },
+        text: "workflow evidence",
+      },
+    ],
+    missingEvidence: [],
+    query: "workflow question",
+    state: "answerable",
+  };
+
+  it.each(["postgres", "tidb"] as const)(
+    "lists workflow-sourced traces for any current reader and reads their embedded evidence on %s",
+    async (dialect) => {
+      const calls: DatabaseExecuteInput[] = [];
+      const database = testDatabase(dialect, async (input) => {
+        calls.push(input);
+        if (input.tableName === "answer_traces") {
+          return {
+            rows: [
+              {
+                completed: dialect === "tidb" ? "1" : true,
+                created_at: NOW,
+                evidence_bundle_id: null,
+                evidence_items: null,
+                evidence_state: null,
+                id: WORKFLOW_TRACE_ID,
+                mode: "fast",
+                query: "workflow question",
+                source: "workflow",
+              },
+            ],
+            rowsAffected: 1,
+          };
+        }
+        if (input.tableName === "answer_trace_steps") {
+          const metadata = { evidenceBundle, resultCount: 1 };
+          return {
+            rows: [
+              {
+                ended_at: NOW,
+                id: "018f0d60-7a49-7cc2-9c1b-5b36f18f2c74",
+                metadata: dialect === "tidb" ? JSON.stringify(metadata) : metadata,
+                name: "retrieval.test",
+                started_at: NOW,
+                status: "ok",
+                trace_id: WORKFLOW_TRACE_ID,
+              },
+            ],
+            rowsAffected: 1,
+          };
+        }
+        return { rows: [], rowsAffected: 0 };
+      });
+      const repository = createDatabaseQualityControlRepository({ database, maxListLimit: 100 });
+
+      const page = await repository.listTraces({
+        candidateGrants: ["tenant:tenant-1"],
+        capabilityRequester: { callerKind: "interactive", subjectId: "reader-1" },
+        knowledgeSpaceId: SPACE_ID,
+        limit: 10,
+        source: "workflow",
+        subjectId: "reader-1",
+        tenantId: "tenant-1",
+      });
+
+      expect(page.items[0]).toMatchObject({
+        evidenceState: "answerable",
+        id: WORKFLOW_TRACE_ID,
+        resultCount: 1,
+        scores: { final: 0.8, retrieval: 0.7 },
+        source: "workflow",
+      });
+      const call = calls.find((candidate) => candidate.tableName === "answer_traces");
+      // Own traces still join through the requester's grant; shared sources need only a current
+      // reader grant on the space, so the join is no longer an inner join.
+      expect(call?.sql).toContain("LEFT JOIN");
+      expect(call?.sql).toContain("'workflow', 'service_api', 'agent', 'mcp'");
+      expect(call?.sql).toContain("reader");
+      expect(call?.sql).not.toContain("knowledge_space_permission_snapshots");
+      expect(call?.params).toContain("workflow");
+    },
+  );
+
+  it.each(["postgres", "tidb"] as const)(
+    "treats rows without a stored source as retrieval tests when filtering on %s",
+    async (dialect) => {
+      const calls: DatabaseExecuteInput[] = [];
+      const database = testDatabase(dialect, async (input) => {
+        calls.push(input);
+        if (input.tableName === "answer_traces") {
+          return {
+            rows: [
+              {
+                completed: dialect === "tidb" ? "1" : true,
+                created_at: NOW,
+                evidence_bundle_id: null,
+                evidence_items: [],
+                evidence_state: null,
+                id: TRACE_ID,
+                mode: "fast",
+                query: "legacy question",
+                source: null,
+              },
+            ],
+            rowsAffected: 1,
+          };
+        }
+        return { rows: [], rowsAffected: 0 };
+      });
+      const repository = createDatabaseQualityControlRepository({ database, maxListLimit: 100 });
+
+      const page = await repository.listTraces({
+        candidateGrants: ["tenant:tenant-1"],
+        knowledgeSpaceId: SPACE_ID,
+        limit: 10,
+        source: "retrieval_test",
+        subjectId: "editor-1",
+        tenantId: "tenant-1",
+      });
+
+      expect(page.items[0]?.source).toBe("retrieval_test");
+      const call = calls.find((candidate) => candidate.tableName === "answer_traces");
+      expect(call?.sql).toMatch(/IS NULL OR trace\.[`"]source[`"] = /u);
+    },
+  );
+});
diff --git a/knowledge-fs/packages/api/src/quality-control-database-repository.ts b/knowledge-fs/packages/api/src/quality-control-database-repository.ts
index 834d2c7c4bb..b29805a8193 100644
--- a/knowledge-fs/packages/api/src/quality-control-database-repository.ts
+++ b/knowledge-fs/packages/api/src/quality-control-database-repository.ts
@@ -6,7 +6,12 @@ import type {
   DatabaseQueryValue,
   DatabaseRow,
 } from "@knowledge/core";
-import { QueryImageMetadataSchema } from "@knowledge/core";
+import {
+  type AnswerTraceSource,
+  AnswerTraceSourceSchema,
+  EvidenceBundleSchema,
+  QueryImageMetadataSchema,
+} from "@knowledge/core";
 
 import { answerTraceKnowledgeSpaceVisibilitySql } from "./answer-trace-repository";
 import { resolveCapabilityJobPublicationGrant } from "./capability-job-fence";
@@ -1080,18 +1085,28 @@ async function listTraces(
   const params: DatabaseQueryValue[] = [input.tenantId, input.knowledgeSpaceId, input.subjectId];
   let authorizationFilter: string;
   let authorizationJoin: string;
+  // A subject sees the traces it produced itself (joined through the grant or permission snapshot
+  // that authorized them) plus every trace another caller kind produced in the space (workflow
+  // nodes, agents, service API, MCP), provided the subject currently holds an authorization for
+  // the space whose content scope admits the caller's candidate grants. Console retrieval tests of
+  // other members stay private to their author, exactly as before.
+  const sharedSourcesSql = SHARED_ANSWER_TRACE_SOURCES.map((source) => `'${source}'`).join(", ");
   if (input.capabilityRequester) {
     params.push(input.capabilityRequester.callerKind, JSON.stringify(input.candidateGrants));
-    authorizationJoin = `INNER JOIN ${q(database, "capability_grants")} capability ON capability.${q(database, "tenant_id")} = ${p(database, 1)} AND capability.${q(database, "knowledge_space_id")} = trace.${q(database, "knowledge_space_id")} AND capability.${q(database, "grant_id")} = trace.${q(database, "capability_grant_id")} AND capability.${q(database, "subject_id")} = ${p(database, 3)} AND capability.${q(database, "caller_kind")} = ${p(database, 4)}`;
-    authorizationFilter = permissionScopeSql(
+    authorizationJoin = `LEFT JOIN ${q(database, "capability_grants")} capability ON capability.${q(database, "tenant_id")} = ${p(database, 1)} AND capability.${q(database, "knowledge_space_id")} = trace.${q(database, "knowledge_space_id")} AND capability.${q(database, "grant_id")} = trace.${q(database, "capability_grant_id")} AND capability.${q(database, "subject_id")} = ${p(database, 3)} AND capability.${q(database, "caller_kind")} = ${p(database, 4)}`;
+    const ownTrace = `capability.${q(database, "grant_id")} IS NOT NULL AND ${permissionScopeSql(
       database,
       `capability.${q(database, "content_scope_ids")}`,
       p(database, 5),
-    );
+    )}`;
+    const sharedTrace = `trace.${q(database, "source")} IN (${sharedSourcesSql}) AND EXISTS (SELECT 1 FROM ${q(database, "capability_grants")} reader WHERE reader.${q(database, "tenant_id")} = ${p(database, 1)} AND reader.${q(database, "knowledge_space_id")} = trace.${q(database, "knowledge_space_id")} AND reader.${q(database, "subject_id")} = ${p(database, 3)} AND reader.${q(database, "caller_kind")} = ${p(database, 4)} AND reader.${q(database, "state")} = 'active' AND ${permissionScopeSql(database, `reader.${q(database, "content_scope_ids")}`, p(database, 5))})`;
+    authorizationFilter = `((${ownTrace}) OR (${sharedTrace}))`;
   } else {
     params.push(JSON.stringify(input.candidateGrants));
-    authorizationJoin = `INNER JOIN ${q(database, "knowledge_space_permission_snapshots")} permission ON permission.${q(database, "tenant_id")} = ${p(database, 1)} AND permission.${q(database, "knowledge_space_id")} = trace.${q(database, "knowledge_space_id")} AND permission.${q(database, "id")} = trace.${q(database, "permission_snapshot_id")} AND permission.${q(database, "subject_id")} = trace.${q(database, "subject_id")}`;
-    authorizationFilter = `trace.${q(database, "subject_id")} = ${p(database, 3)} AND ${permissionScopeSql(database, `permission.${q(database, "permission_scopes")}`, p(database, 4))}`;
+    authorizationJoin = `LEFT JOIN ${q(database, "knowledge_space_permission_snapshots")} permission ON permission.${q(database, "tenant_id")} = ${p(database, 1)} AND permission.${q(database, "knowledge_space_id")} = trace.${q(database, "knowledge_space_id")} AND permission.${q(database, "id")} = trace.${q(database, "permission_snapshot_id")} AND permission.${q(database, "subject_id")} = trace.${q(database, "subject_id")}`;
+    const ownTrace = `trace.${q(database, "subject_id")} = ${p(database, 3)} AND permission.${q(database, "id")} IS NOT NULL AND ${permissionScopeSql(database, `permission.${q(database, "permission_scopes")}`, p(database, 4))}`;
+    const sharedTrace = `trace.${q(database, "source")} IN (${sharedSourcesSql}) AND EXISTS (SELECT 1 FROM ${q(database, "knowledge_space_permission_snapshots")} reader WHERE reader.${q(database, "tenant_id")} = ${p(database, 1)} AND reader.${q(database, "knowledge_space_id")} = trace.${q(database, "knowledge_space_id")} AND reader.${q(database, "subject_id")} = ${p(database, 3)} AND ${permissionScopeSql(database, `reader.${q(database, "permission_scopes")}`, p(database, 4))})`;
+    authorizationFilter = `((${ownTrace}) OR (${sharedTrace}))`;
   }
   const traceTenantFilter = input.capabilityRequester
     ? `trace.${q(database, "tenant_id")} = ${p(database, 1)} AND `
@@ -1115,6 +1130,14 @@ async function listTraces(
     params.push(input.mode);
     filters.push(`trace.${q(database, "mode")} = ${p(database, params.length)}`);
   }
+  if (input.source) {
+    params.push(input.source);
+    filters.push(
+      input.source === "retrieval_test"
+        ? `(trace.${q(database, "source")} IS NULL OR trace.${q(database, "source")} = ${p(database, params.length)})`
+        : `trace.${q(database, "source")} = ${p(database, params.length)}`,
+    );
+  }
   if (input.status) {
     filters.push(
       input.status === "completed"
@@ -1176,7 +1199,13 @@ function mapTraceSummary(
   row: DatabaseRow,
   steps: readonly DatabaseRow[],
 ): QualityAnswerTraceSummary {
-  const evidenceItems = row.evidence_items == null ? [] : jsonArrayColumn(row, "evidence_items");
+  // Retrieval tests recorded by workflow nodes carry their evidence inside the summary step instead
+  // of a persisted bundle row; read counts and scores from there when no bundle is joined.
+  const embeddedBundle = row.evidence_items == null ? embeddedEvidenceBundle(steps) : undefined;
+  const evidenceItems =
+    row.evidence_items == null
+      ? (embeddedBundle?.items ?? [])
+      : jsonArrayColumn(row, "evidence_items");
   const queryImages =
     row.query_images == null
       ? []
@@ -1218,12 +1247,13 @@ function mapTraceSummary(
     ...(optionalStringColumn(row, "evidence_bundle_id")
       ? { evidenceBundleId: optionalStringColumn(row, "evidence_bundle_id") }
       : {}),
-    ...(optionalStringColumn(row, "evidence_state")
-      ? { evidenceState: optionalStringColumn(row, "evidence_state") }
+    ...((optionalStringColumn(row, "evidence_state") ?? embeddedBundle?.state)
+      ? { evidenceState: optionalStringColumn(row, "evidence_state") ?? embeddedBundle?.state }
       : {}),
     ...(score("final") !== undefined ? { finalScore: score("final") } : {}),
     id: stringColumn(row, "id"),
     mode: stringColumn(row, "mode") as QualityAnswerTraceSummary["mode"],
+    source: traceSourceColumn(row),
     ...(optionalStringColumn(row, "open_bad_case_id")
       ? { openBadCaseId: optionalStringColumn(row, "open_bad_case_id") }
       : {}),
@@ -2226,3 +2256,22 @@ function p(database: DatabaseAdapter, position: number) {
 function jsonP(database: DatabaseAdapter, position: number) {
   return jsonInsertPlaceholder(database, position, undefined);
 }
+
+/** Every source except console retrieval tests; visible to any current reader of the space. */
+const SHARED_ANSWER_TRACE_SOURCES: readonly AnswerTraceSource[] =
+  AnswerTraceSourceSchema.options.filter((source) => source !== "retrieval_test");
+
+function traceSourceColumn(row: DatabaseRow): AnswerTraceSource {
+  const parsed = AnswerTraceSourceSchema.safeParse(optionalStringColumn(row, "source"));
+  return parsed.success ? parsed.data : "retrieval_test";
+}
+
+function embeddedEvidenceBundle(steps: readonly DatabaseRow[]) {
+  for (const step of steps) {
+    const candidate = EvidenceBundleSchema.safeParse(
+      jsonObjectColumn(step, "metadata").evidenceBundle,
+    );
+    if (candidate.success) return candidate.data;
+  }
+  return undefined;
+}
diff --git a/knowledge-fs/packages/api/src/quality-control-handlers.ts b/knowledge-fs/packages/api/src/quality-control-handlers.ts
index 22a54745684..4cfb97c32d0 100644
--- a/knowledge-fs/packages/api/src/quality-control-handlers.ts
+++ b/knowledge-fs/packages/api/src/quality-control-handlers.ts
@@ -108,6 +108,7 @@ export function registerQualityControlHandlers({
         limit: query.limit,
         ...(query.mode ? { mode: query.mode } : {}),
         ...(query.query ? { query: query.query } : {}),
+        ...(query.source ? { source: query.source } : {}),
         ...(query.status ? { status: query.status } : {}),
         subjectId: scope.subject.subjectId,
         tenantId: scope.subject.tenantId,
diff --git a/knowledge-fs/packages/api/src/quality-control-routes.ts b/knowledge-fs/packages/api/src/quality-control-routes.ts
index 425d4617759..3b1b16f689a 100644
--- a/knowledge-fs/packages/api/src/quality-control-routes.ts
+++ b/knowledge-fs/packages/api/src/quality-control-routes.ts
@@ -1,5 +1,5 @@
 import { createRoute, z } from "@hono/zod-openapi";
-import { QueryImageMetadataSchema } from "@knowledge/core";
+import { AnswerTraceSourceSchema, QueryImageMetadataSchema } from "@knowledge/core";
 
 import { ForbiddenResponse, UnauthorizedResponse } from "./gateway-openapi-contracts";
 import { ErrorResponseSchema } from "./gateway-route-schemas";
@@ -28,6 +28,7 @@ export const QualityTraceHistoryQuerySchema = z
     limit: z.coerce.number().int().min(1).max(100).default(50),
     mode: z.enum(["auto", "deep", "fast", "research"]).optional(),
     query: z.string().trim().min(1).max(500).optional(),
+    source: AnswerTraceSourceSchema.optional(),
     status: z.enum(["completed", "failed"]).optional(),
     to: DateTime.optional(),
   })
@@ -58,6 +59,7 @@ export const QualityTraceSummarySchema = z
     queryImages: z.array(QueryImageMetadataSchema).max(4).optional(),
     resultCount: z.number().int().nonnegative(),
     scores: z.object({ final: ScoreSchema, rerank: ScoreSchema, retrieval: ScoreSchema }),
+    source: AnswerTraceSourceSchema,
     stages: z.array(
       z.object({
         candidateCount: z.number().int().nonnegative().optional(),
diff --git a/knowledge-fs/packages/api/src/quality-control.ts b/knowledge-fs/packages/api/src/quality-control.ts
index 8914fe3355c..c4322b84dcc 100644
--- a/knowledge-fs/packages/api/src/quality-control.ts
+++ b/knowledge-fs/packages/api/src/quality-control.ts
@@ -1,6 +1,7 @@
 import { randomUUID } from "node:crypto";
 
 import type {
+  AnswerTraceSource,
   AuthSubject,
   KnowledgeSpaceEmbeddingProfile,
   KnowledgeSpaceRetrievalProfile,
@@ -55,6 +56,8 @@ export interface QualityAnswerTraceSummary {
   readonly finalScore?: number | undefined;
   readonly id: string;
   readonly mode: "auto" | "deep" | "fast" | "research";
+  /** Caller that produced the trace; legacy rows without a stored value are retrieval tests. */
+  readonly source: AnswerTraceSource;
   /** The unresolved (open or replaying) bad case already filed for this trace, if any. */
   readonly openBadCaseId?: string | undefined;
   readonly profile: {
@@ -95,6 +98,7 @@ export interface QualityAnswerTraceHistoryInput {
   readonly knowledgeSpaceId: string;
   readonly limit: number;
   readonly mode?: "auto" | "deep" | "fast" | "research" | undefined;
+  readonly source?: AnswerTraceSource | undefined;
   readonly query?: string | undefined;
   readonly status?: "completed" | "failed" | undefined;
   readonly subjectId: string;
diff --git a/knowledge-fs/packages/api/src/query-handlers.ts b/knowledge-fs/packages/api/src/query-handlers.ts
index f3a8a13336e..87b1a7f3a5c 100644
--- a/knowledge-fs/packages/api/src/query-handlers.ts
+++ b/knowledge-fs/packages/api/src/query-handlers.ts
@@ -2,6 +2,7 @@ import type { OpenAPIHono } from "@hono/zod-openapi";
 import { validateKnowledgeSpaceRetrievalProfileForMode } from "@knowledge/core";
 
 import type { AnswerTraceRecorder } from "./answer-trace-recorder";
+import { answerTraceSourceForCaller } from "./answer-trace-source";
 import {
   type AutoRetrievalModeResolver,
   resolveRetrievalModeRequest,
@@ -500,6 +501,10 @@ export function registerQueryHandlers({
         input: {
           ...(capabilityGrant ? { capabilityGrantId: capabilityGrant.grantId } : {}),
           knowledgeSpaceId: space.id,
+          source: answerTraceSourceForCaller({
+            callerKind: context.get("callerKind"),
+            capabilityCallerKind: capabilityGrant?.callerKind,
+          }),
           ...(runtimeSnapshot?.embeddingProfile
             ? { embeddingProfile: runtimeSnapshot.embeddingProfile }
             : {}),
diff --git a/knowledge-fs/packages/api/src/research-task-handlers.test.ts b/knowledge-fs/packages/api/src/research-task-handlers.test.ts
index 5db0c7725f1..5273fa839c1 100644
--- a/knowledge-fs/packages/api/src/research-task-handlers.test.ts
+++ b/knowledge-fs/packages/api/src/research-task-handlers.test.ts
@@ -15,6 +15,7 @@ import {
   encodeResearchTaskListCursor,
   researchTaskRuntimeSnapshotFromMetadata,
 } from "./index";
+import { createInMemoryKnowledgeSpaceOverviewRepository } from "./knowledge-space-overview";
 
 describe("research task handlers", () => {
   it("round-trips list cursors and rejects every malformed cursor boundary", () => {
@@ -84,9 +85,14 @@ describe("research task handlers", () => {
       repository: createInMemoryResearchTaskJobRepository({ maxJobs: 10 }),
     });
     let deletionActive = false;
+    const overview = createInMemoryKnowledgeSpaceOverviewRepository({
+      maxEvents: 100,
+      maxListLimit: 50,
+    });
     const app = createKnowledgeGateway({
       adapter,
       allowLegacyResearchTaskProfileFallback: true,
+      knowledgeSpaceOverview: overview,
       auth: createStaticAuthVerifier({
         subjectsByToken: {
           "editor-token": {
@@ -129,6 +135,27 @@ describe("research task handlers", () => {
     });
 
     expect(response.status).toBe(201);
+    // Research tasks count as knowledge-space queries in the overview, keyed by the task id.
+    expect(
+      (
+        await overview.listActivity({
+          candidateGrants: ["tenant:tenant-1"],
+          knowledgeSpaceId,
+          limit: 10,
+          tenantId: "tenant-1",
+        })
+      ).items,
+    ).toEqual([
+      expect.objectContaining({
+        action: "query.requested",
+        details: expect.objectContaining({
+          question: "Research semantic retrieval regressions",
+          source: "retrieval_test",
+          taskKind: "research",
+        }),
+        resource: { id: "research-task-job-1", type: "query" },
+      }),
+    ]);
     const responseBody = (await response.json()) as Record<string, unknown>;
     expect(responseBody).toMatchObject({
       id: "research-task-job-1",
diff --git a/knowledge-fs/packages/api/src/research-task-handlers.ts b/knowledge-fs/packages/api/src/research-task-handlers.ts
index dd729ce6f5f..e52eab5dd83 100644
--- a/knowledge-fs/packages/api/src/research-task-handlers.ts
+++ b/knowledge-fs/packages/api/src/research-task-handlers.ts
@@ -4,6 +4,7 @@ import {
   validateKnowledgeSpaceRetrievalProfileForMode,
 } from "@knowledge/core";
 
+import { answerTraceSourceForCaller } from "./answer-trace-source";
 import { isAuthenticatedApiKeyBoundToKnowledgeSpace } from "./auth";
 import {
   AUTO_RETRIEVAL_MODE_DECISION_METADATA_KEY,
@@ -31,6 +32,10 @@ import {
   type KnowledgeSpaceAuthorizationGuard,
   type KnowledgeSpaceCallerKind,
 } from "./knowledge-space-authorization";
+import {
+  type KnowledgeSpaceOverviewRepository,
+  deterministicKnowledgeSpaceActivityId,
+} from "./knowledge-space-overview";
 import type { KnowledgeSpaceRepository } from "./knowledge-space-repository";
 import { type LooseOpenApiContext, openApiHandler } from "./openapi-handler-utils";
 import type { PublishedKnowledgeSpaceRuntimeSnapshotResolver } from "./published-knowledge-space-runtime-snapshot";
@@ -88,6 +93,8 @@ export interface RegisterResearchTaskHandlersOptions {
   readonly assets: Pick<DocumentAssetRepository, "get" | "getManyByIds">;
   readonly dryRunResearchPlanner: ResearchTaskDryRunPlanner;
   readonly deletionVisibility?: ResearchTaskDeletionVisibility | undefined;
+  /** Counts each task as a knowledge-space query in the overview, like the query stream does. */
+  readonly overview?: Pick<KnowledgeSpaceOverviewRepository, "appendActivity"> | undefined;
   readonly researchTaskJobs: ResearchTaskJobStateMachine;
   readonly researchTaskPartialResults: ResearchTaskPartialResultRepository;
   readonly researchTaskProgressEvents: ResearchTaskProgressRepository;
@@ -109,6 +116,7 @@ export function registerResearchTaskHandlers({
   dryRunResearchPlanner,
   deletionVisibility,
   directStream,
+  overview,
   researchTaskJobs,
   researchTaskPartialResults,
   researchTaskProgressEvents,
@@ -390,6 +398,17 @@ export function registerResearchTaskHandlers({
           tenantId: subject.tenantId,
           topK: plan.retrievalPlan.topK,
         });
+        await appendResearchTaskQueryRequested({
+          job,
+          occurredAt: new Date(now()).toISOString(),
+          overview,
+          query: body.query,
+          source: answerTraceSourceForCaller({
+            callerKind,
+            capabilityCallerKind: capabilityGrant?.callerKind,
+          }),
+          subject,
+        });
 
         return context.json(toPublicResearchTaskJob(job), 201);
       } catch (error) {
@@ -927,3 +946,58 @@ async function authorizeJobAccess(input: {
     throw error;
   }
 }
+
+/**
+ * Count the Research task as a knowledge-space query. The overview derives query volume from
+ * `query.requested` events and marks the task answered when the job state machine emits
+ * `query.completed` for the same id. Best effort: task creation must not fail because the
+ * activity feed is unavailable.
+ */
+async function appendResearchTaskQueryRequested({
+  job,
+  occurredAt,
+  overview,
+  query,
+  source,
+  subject,
+}: {
+  readonly job: {
+    readonly id: string;
+    readonly knowledgeSpaceId: string;
+    readonly mode?: string | undefined;
+    readonly tenantId: string;
+  };
+  readonly occurredAt: string;
+  readonly overview: Pick<KnowledgeSpaceOverviewRepository, "appendActivity"> | undefined;
+  readonly query: string | undefined;
+  readonly source: string;
+  readonly subject: KnowledgeGatewayEnv["Variables"]["subject"];
+}): Promise<void> {
+  if (!overview) return;
+  try {
+    await overview.appendActivity({
+      action: "query.requested",
+      actor: { id: subject.subjectId, type: "member" },
+      details: {
+        mode: job.mode ?? "research",
+        ...(query ? { question: query } : {}),
+        source,
+        taskKind: "research",
+      },
+      id: deterministicKnowledgeSpaceActivityId(
+        "query.requested",
+        job.tenantId,
+        job.knowledgeSpaceId,
+        job.id,
+      ),
+      knowledgeSpaceId: job.knowledgeSpaceId,
+      occurredAt,
+      requiredPermissionScope: [],
+      resource: { id: job.id, type: "query" },
+      result: "success",
+      tenantId: job.tenantId,
+    });
+  } catch {
+    // Best effort: see the function comment.
+  }
+}
diff --git a/knowledge-fs/packages/api/src/research-task-job.test.ts b/knowledge-fs/packages/api/src/research-task-job.test.ts
index 6d15395702d..f55b6066d0f 100644
--- a/knowledge-fs/packages/api/src/research-task-job.test.ts
+++ b/knowledge-fs/packages/api/src/research-task-job.test.ts
@@ -1009,3 +1009,74 @@ function evidenceBundle(id: string, text: string) {
     state: "partial" as const,
   };
 }
+
+describe("research task overview outcomes", () => {
+  it("reports completion and failure to the space overview as query outcomes", async () => {
+    const appendActivity = vi.fn(async (input: unknown) => input as never);
+    const machine = createResearchTaskJobStateMachine({
+      generateId: (() => {
+        let next = 0;
+        return () => `research-task-job-${++next}`;
+      })(),
+      jobs: new FakeJobQueue(),
+      now: () => 5_000,
+      overview: { appendActivity },
+      repository: createInMemoryResearchTaskJobRepository({ maxJobs: 10 }),
+    });
+
+    const completed = await machine.start(baseStartInput());
+    for (const stage of [
+      "planning",
+      "retrieving",
+      "analyzing",
+      "generating",
+      "completed",
+    ] as const) {
+      await machine.advance(completed.id, stage);
+    }
+    const failed = await machine.start(baseStartInput());
+    await machine.fail(failed.id, "model unavailable");
+    const canceled = await machine.start(baseStartInput());
+    await machine.cancel(canceled.id, "operator");
+
+    expect(appendActivity.mock.calls.map((call) => call[0])).toEqual([
+      expect.objectContaining({
+        action: "query.completed",
+        actor: { type: "system" },
+        knowledgeSpaceId: "space-1",
+        resource: { id: completed.id, type: "query" },
+        result: "success",
+        tenantId: "tenant-1",
+      }),
+      expect.objectContaining({
+        action: "query.failed",
+        details: expect.objectContaining({ reasonCode: "failed", taskKind: "research" }),
+        resource: { id: failed.id, type: "query" },
+        result: "failure",
+      }),
+      expect.objectContaining({
+        action: "query.failed",
+        details: expect.objectContaining({ reasonCode: "canceled", taskKind: "research" }),
+        resource: { id: canceled.id, type: "query" },
+        result: "canceled",
+      }),
+    ]);
+  });
+
+  it("never lets an overview write failure break a terminal transition", async () => {
+    const machine = createResearchTaskJobStateMachine({
+      generateId: () => "research-task-job-1",
+      jobs: new FakeJobQueue(),
+      now: () => 5_000,
+      overview: {
+        appendActivity: async () => {
+          throw new Error("activity feed down");
+        },
+      },
+      repository: createInMemoryResearchTaskJobRepository({ maxJobs: 10 }),
+    });
+    const job = await machine.start(baseStartInput());
+
+    await expect(machine.fail(job.id, "boom")).resolves.toMatchObject({ stage: "failed" });
+  });
+});
diff --git a/knowledge-fs/packages/api/src/research-task-job.ts b/knowledge-fs/packages/api/src/research-task-job.ts
index 3390384c9c4..6d8453f419e 100644
--- a/knowledge-fs/packages/api/src/research-task-job.ts
+++ b/knowledge-fs/packages/api/src/research-task-job.ts
@@ -4,6 +4,10 @@ import {
   type JobPayload,
   type JobQueueAdapter,
 } from "@knowledge/core";
+import {
+  type KnowledgeSpaceOverviewRepository,
+  deterministicKnowledgeSpaceActivityId,
+} from "./knowledge-space-overview";
 
 import type {
   CapabilityCallerKind,
@@ -196,6 +200,8 @@ export interface ResearchTaskJobStateMachineOptions {
   readonly maxQueryBytes?: number;
   readonly metrics?: DurableTaskOperationalMetrics | undefined;
   readonly now?: () => number;
+  /** Reports terminal outcomes to the space overview so Research tasks count as answered/failed queries. */
+  readonly overview?: Pick<KnowledgeSpaceOverviewRepository, "appendActivity"> | undefined;
   readonly progress?: ResearchTaskProgressPublisher | undefined;
   readonly repository: ResearchTaskJobRepository;
 }
@@ -291,6 +297,7 @@ export function createResearchTaskJobStateMachine({
   maxQueryBytes = defaultMaxQueryBytes,
   metrics,
   now = Date.now,
+  overview,
   progress,
   repository,
 }: ResearchTaskJobStateMachineOptions): ResearchTaskJobStateMachine {
@@ -316,6 +323,7 @@ export function createResearchTaskJobStateMachine({
           outcome: "completed",
           taskKind: "research",
         });
+        await appendResearchTaskOutcome(overview, updated, "query.completed", "success", {});
       }
       await progress?.publish(updated, "research_task.stage_changed", {
         previousStage: job.stage,
@@ -341,6 +349,9 @@ export function createResearchTaskJobStateMachine({
         outcome: "canceled",
         taskKind: "research",
       });
+      await appendResearchTaskOutcome(overview, updated, "query.failed", "canceled", {
+        reasonCode: "canceled",
+      });
       await progress?.publish(updated, "research_task.canceled", reason ? { reason } : {});
       return cloneResearchTaskJob(updated);
     },
@@ -363,6 +374,9 @@ export function createResearchTaskJobStateMachine({
         outcome: "failed",
         taskKind: "research",
       });
+      await appendResearchTaskOutcome(overview, updated, "query.failed", "failure", {
+        reasonCode: "failed",
+      });
       await progress?.publish(updated, "research_task.failed", { error });
       return cloneResearchTaskJob(updated);
     },
@@ -1135,3 +1149,34 @@ function clonePartialResult(result: ResearchTaskPartialResult): ResearchTaskPart
 function cloneEvidenceBundle(bundle: EvidenceBundle): EvidenceBundle {
   return JSON.parse(JSON.stringify(bundle)) as EvidenceBundle;
 }
+
+/**
+ * Mirror a terminal Research outcome into the space overview. The task was counted as a query
+ * when it was created; `query.completed` marks it answered and `query.failed` keeps it in the
+ * unanswered bucket. Best effort: the state transition has already been persisted.
+ */
+async function appendResearchTaskOutcome(
+  overview: Pick<KnowledgeSpaceOverviewRepository, "appendActivity"> | undefined,
+  job: ResearchTaskJob,
+  action: "query.completed" | "query.failed",
+  result: "canceled" | "failure" | "success",
+  details: Readonly<Record<string, unknown>>,
+): Promise<void> {
+  if (!overview) return;
+  try {
+    await overview.appendActivity({
+      action,
+      actor: { type: "system" },
+      details: { ...details, taskKind: "research" },
+      id: deterministicKnowledgeSpaceActivityId(action, job.tenantId, job.knowledgeSpaceId, job.id),
+      knowledgeSpaceId: job.knowledgeSpaceId,
+      occurredAt: new Date(job.completedAt ?? job.updatedAt).toISOString(),
+      requiredPermissionScope: [],
+      resource: { id: job.id, type: "query" },
+      result,
+      tenantId: job.tenantId,
+    });
+  } catch {
+    // Best effort: see the function comment.
+  }
+}
diff --git a/knowledge-fs/packages/api/src/retrieval-test-handlers.test.ts b/knowledge-fs/packages/api/src/retrieval-test-handlers.test.ts
index bcc9b421fc9..b47eea4c07d 100644
--- a/knowledge-fs/packages/api/src/retrieval-test-handlers.test.ts
+++ b/knowledge-fs/packages/api/src/retrieval-test-handlers.test.ts
@@ -90,21 +90,28 @@ describe("retrieval test route", () => {
       });
 
       expect(response.status).toBe(200);
+      // The node forwards this id when it captures an empty retrieval, so one record stays per run.
+      expect((await response.json()) as Record<string, unknown>).toMatchObject({
+        answerTraceId: expectedTraceId,
+      });
       expect(overview.appendActivity).toHaveBeenCalledWith(
         expect.objectContaining({
           action: "query.requested",
-          details: expect.objectContaining({ mode: "fast" }),
+          details: expect.objectContaining({ mode: "fast", source: "workflow" }),
           resource: { id: expectedTraceId, type: "query" },
         }),
       );
       expect(answerTraceRecorder.record).toHaveBeenCalledWith(
         expect.objectContaining({
           capabilityGrantId: "10000000-0000-4000-8000-000000000011",
+          source: "workflow",
           traceId: expectedTraceId,
           steps: [
             expect.objectContaining({
               metadata: expect.objectContaining({
                 queryOutcome: expectedOutcome,
+                resultCount: result.items.length,
+                stages: result.stages,
                 workflowQueryId: WORKFLOW_QUERY_ID,
               }),
               name: "query.generate",
diff --git a/knowledge-fs/packages/api/src/retrieval-test-handlers.ts b/knowledge-fs/packages/api/src/retrieval-test-handlers.ts
index 91d2808e244..0a261708ff4 100644
--- a/knowledge-fs/packages/api/src/retrieval-test-handlers.ts
+++ b/knowledge-fs/packages/api/src/retrieval-test-handlers.ts
@@ -2,10 +2,14 @@ import type { OpenAPIHono } from "@hono/zod-openapi";
 import { validateKnowledgeSpaceRetrievalProfileForMode } from "@knowledge/core";
 
 import type { AnswerTraceRecorder } from "./answer-trace-recorder";
+import { answerTraceSourceForCaller } from "./answer-trace-source";
 import { currentCandidateGrants } from "./candidate-content-authorization";
 import { classifyQueryOutcome } from "./failed-query-recorder";
 import type { KnowledgeGatewayEnv } from "./gateway-openapi-contracts";
-import { queryRetrievalProfileMetadata } from "./gateway-sse-responses";
+import {
+  queryProjectionSnapshotMetadata,
+  queryRetrievalProfileMetadata,
+} from "./gateway-sse-responses";
 import {
   type KnowledgeSpaceOverviewRepository,
   deterministicKnowledgeSpaceActivityId,
@@ -70,6 +74,10 @@ export function registerRetrievalTestHandlers({
     const body = context.req.valid("json");
     const capabilityGrant = context.get("capabilityV2Grant");
     const workflowQueryId = capabilityGrant?.callerKind === "workflow" ? body.queryId : undefined;
+    const source = answerTraceSourceForCaller({
+      callerKind: context.get("callerKind"),
+      capabilityCallerKind: capabilityGrant?.callerKind,
+    });
     const space = await spaces.get({ id: knowledgeSpaceId, tenantId: subject.tenantId });
     if (!space) {
       return context.json({ error: "Knowledge space not found" }, 404);
@@ -233,6 +241,7 @@ export function registerRetrievalTestHandlers({
             details: {
               mode,
               ...(body.query ? { question: body.query } : {}),
+              source,
             },
             id: deterministicKnowledgeSpaceActivityId(
               "query.requested",
@@ -274,6 +283,9 @@ export function registerRetrievalTestHandlers({
           traceId,
         });
         await executionLease.assertActive();
+        const recordsTrace = Boolean(
+          workflowAnswerTraceId && capabilityGrant && answerTraceRecorder && overview,
+        );
         if (workflowAnswerTraceId && capabilityGrant && answerTraceRecorder && overview) {
           const finishReason =
             result.items.length > 0 ? "retrieval-evidence" : "no-retrieval-evidence";
@@ -297,6 +309,9 @@ export function registerRetrievalTestHandlers({
                 }
               : {}),
           });
+          // One `query.generate` step carries the outcome classification the overview reads plus
+          // everything the retrieval history needs to show the run like a console retrieval test:
+          // the evidence bundle (counts, scores, evidence view), stages, plan, and publication.
           await answerTraceRecorder.record({
             capabilityGrantId: capabilityGrant.grantId,
             knowledgeSpaceId,
@@ -305,12 +320,20 @@ export function registerRetrievalTestHandlers({
             ...(resolvedQueryImages.length > 0
               ? { queryImages: resolvedQueryImages.map(queryImageMetadata) }
               : {}),
+            source,
             steps: [
               {
                 metadata: {
                   ...outcomeMetadata,
+                  ...(result.evidenceBundle ? { evidenceBundle: result.evidenceBundle } : {}),
+                  plan: result.plan,
+                  projectionSnapshot: queryProjectionSnapshotMetadata(
+                    runtimeSnapshot.projectionSnapshot,
+                  ),
                   queryOutcome: classification.outcome,
                   ...(classification.trigger ? { failedQueryTrigger: classification.trigger } : {}),
+                  resultCount: result.items.length,
+                  stages: result.stages,
                 },
                 name: "query.generate",
                 status: "ok",
@@ -327,6 +350,10 @@ export function registerRetrievalTestHandlers({
           : "disabled";
 
         const response = RetrievalTestResponseSchema.parse({
+          // Lets the caller attach a later failed-retrieval capture to this history record.
+          ...(recordsTrace && workflowAnswerTraceId
+            ? { answerTraceId: workflowAnswerTraceId }
+            : {}),
           capabilityStatus: {
             embedding: embeddingCapabilityStatus,
             reasoning: "verified" as const,
diff --git a/knowledge-fs/packages/api/src/retrieval-test-routes.ts b/knowledge-fs/packages/api/src/retrieval-test-routes.ts
index 76c4e3fabd7..87a7b23a809 100644
--- a/knowledge-fs/packages/api/src/retrieval-test-routes.ts
+++ b/knowledge-fs/packages/api/src/retrieval-test-routes.ts
@@ -204,6 +204,8 @@ export const RetrievalTestMetricsSchema = z
 
 export const RetrievalTestResponseSchema = z
   .object({
+    /** AnswerTrace recorded for this run in the space's retrieval history, when one was written. */
+    answerTraceId: z.string().uuid().optional(),
     capabilityStatus: z
       .object({
         embedding: z.enum(["not-required", "verified"]),
diff --git a/knowledge-fs/packages/api/src/retrieval-test.test.ts b/knowledge-fs/packages/api/src/retrieval-test.test.ts
index 33350d8e942..0a79c545973 100644
--- a/knowledge-fs/packages/api/src/retrieval-test.test.ts
+++ b/knowledge-fs/packages/api/src/retrieval-test.test.ts
@@ -231,6 +231,9 @@ describe("createRetrievalTestExecutor", () => {
         sources: ["dense", "fts"],
       },
     ]);
+    // This fixture's node/document ids are not UUIDs, so no history bundle can be assembled; the
+    // diagnostics are unaffected. See the dedicated evidence bundle test for the real shape.
+    expect(result.evidenceBundle).toBeUndefined();
     expect(JSON.stringify(result)).not.toContain("secret candidate text");
     expect(stageStatuses(result)).toMatchObject({
       dense: "executed",
@@ -734,3 +737,78 @@ function stageStatuses(
 ) {
   return Object.fromEntries(result.stages.map((stage) => [stage.name, stage.status]));
 }
+
+describe("retrieval test history evidence", () => {
+  it("assembles the same EvidenceBundle a query would produce from the raw retrieval", async () => {
+    const nodeId = "018f0d60-7a49-7cc2-9c1b-5b36f18f2c61";
+    const documentAssetId = "018f0d60-7a49-7cc2-9c1b-5b36f18f2c62";
+    const embeddings: EmbeddingProvider = {
+      embed: async () => ({
+        dense: [[0.1, 0.2, 0.3]],
+        metadata: { dimension: 3, model: embeddingSelection.model, provider: "dify-model-runtime" },
+        model: embeddingSelection.model,
+      }),
+      kind: "dify-model-runtime",
+      models: async () => [],
+    };
+    const planner = createRetrievalPlanner({ maxTopK: 100 });
+    const retriever: BasicHybridRetriever = {
+      retrieve: async (input) => ({
+        items: [
+          {
+            citation: {
+              artifactHash: "a".repeat(64),
+              documentAssetId,
+              documentVersion: 1,
+              sectionPath: ["Sensor"],
+            },
+            metadata: { text: "camera sensor evidence text" },
+            nodeId,
+            permissionScope: ["tenant:tenant-1"],
+            projectionIds: ["projection-1"],
+            score: 0.8,
+            sources: ["dense", "fts"],
+          },
+        ],
+        metrics: ordinaryMetrics({ rerank: true }),
+        plan: planner.plan({
+          hasQueryImages: false,
+          mode: "fast",
+          query: input.query,
+          topK: input.topK,
+        }),
+      }),
+    };
+    const executor = createRetrievalTestExecutor({
+      embeddingModel: embeddingSelection.model,
+      embeddings,
+      retriever,
+    });
+
+    const result = await executor.execute({
+      embeddingProfile,
+      // Text is withheld from the diagnostics, but the history bundle still carries evidence.
+      includeText: false,
+      knowledgeSpaceId: SPACE_ID,
+      mode: "fast",
+      permissionScope: ["tenant:tenant-1", "subject:owner-1"],
+      projectionSnapshot,
+      query: "camera sensor evidence",
+      retrievalProfile,
+      subject,
+      traceId: "trace-history",
+    });
+
+    expect(result.items[0]?.text).toBeUndefined();
+    expect(result.evidenceBundle).toMatchObject({
+      query: "camera sensor evidence",
+      state: expect.any(String),
+    });
+    expect(result.evidenceBundle?.items).toHaveLength(1);
+    expect(result.evidenceBundle?.items[0]).toMatchObject({
+      citations: [expect.objectContaining({ documentAssetId })],
+      nodeId,
+      text: "camera sensor evidence text",
+    });
+  });
+});
diff --git a/knowledge-fs/packages/api/src/retrieval-test.ts b/knowledge-fs/packages/api/src/retrieval-test.ts
index 2fb6c662823..d9dc10391a8 100644
--- a/knowledge-fs/packages/api/src/retrieval-test.ts
+++ b/knowledge-fs/packages/api/src/retrieval-test.ts
@@ -1,5 +1,6 @@
 import {
   type AuthSubject,
+  type EvidenceBundle,
   type KnowledgeSpaceEmbeddingProfile,
   type KnowledgeSpaceModelSelection,
   type KnowledgeSpaceRetrievalProfile,
@@ -9,6 +10,7 @@ import type { EmbeddingProvider } from "@knowledge/embeddings";
 
 import { runWithAbortSignal } from "./bounded-concurrency";
 import { candidatePermissionScopeAllows } from "./candidate-content-authorization";
+import { createEvidenceBundleAssembler } from "./evidence-bundle-assembler";
 import {
   type KnowledgeSpaceEmbeddingResolver,
   assertEmbeddingModelMatchesProfile,
@@ -30,6 +32,7 @@ import {
   QUERY_IMAGE_MAX_COUNT,
   QueryImageDegradationReasonSchema,
   type ResolvedQueryImage,
+  queryImageMetadata,
 } from "./query-images";
 import type { RetrievalMetadataFilters } from "./retrieval-candidates";
 import type { RetrievalSource } from "./retrieval-candidates";
@@ -69,6 +72,12 @@ export interface RetrievalTestStage {
 }
 
 export interface RetrievalTestResult {
+  /**
+   * The same EvidenceBundle a query would have produced from this retrieval, so the run can be
+   * recorded as an AnswerTrace with evidence for the retrieval history. Absent when the bundle
+   * could not be assembled; the diagnostics above are unaffected.
+   */
+  readonly evidenceBundle?: EvidenceBundle | undefined;
   readonly items: readonly {
     readonly citation: {
       readonly artifactHash: string;
@@ -294,7 +303,13 @@ export function createRetrievalTestExecutor({
           permissionScope: input.permissionScope,
           profile: input.retrievalProfile,
         });
+        const evidenceBundle = assembleRetrievalTestEvidenceBundle({
+          query: preparedQuery.query || input.query,
+          queryImages: preparedQuery.queryImages,
+          retrieval,
+        });
         return {
+          ...(evidenceBundle ? { evidenceBundle } : {}),
           items: retrieval.items.map((item) =>
             safeRetrievalTestItem(item, input.includeText === true),
           ),
@@ -391,6 +406,37 @@ async function resolveRetrievalTestEmbedding({
   return [...vector];
 }
 
+const retrievalTestEvidenceBundleAssembler = createEvidenceBundleAssembler();
+
+/**
+ * Build the history EvidenceBundle for a retrieval test. Assembly validates every evidence item,
+ * so a retrieval whose items cannot be projected (for example text-less legacy nodes) yields no
+ * bundle rather than failing the diagnostics the caller asked for.
+ */
+function assembleRetrievalTestEvidenceBundle({
+  query,
+  queryImages,
+  retrieval,
+}: {
+  readonly query: string;
+  readonly queryImages: readonly ResolvedQueryImage[];
+  readonly retrieval: Parameters<EvidenceBundleAssemblerAssemble>[0]["retrieval"];
+}): EvidenceBundle | undefined {
+  const images = queryImages.map(queryImageMetadata);
+  if (!query.trim() && images.length === 0) return undefined;
+  try {
+    return retrievalTestEvidenceBundleAssembler.assemble({
+      query,
+      ...(images.length > 0 ? { queryImages: images } : {}),
+      retrieval,
+    });
+  } catch {
+    return undefined;
+  }
+}
+
+type EvidenceBundleAssemblerAssemble = ReturnType<typeof createEvidenceBundleAssembler>["assemble"];
+
 async function prepareRetrievalTestQuery({
   input,
   provider,
diff --git a/knowledge-fs/packages/api/src/workflow-failed-retrieval.test.ts b/knowledge-fs/packages/api/src/workflow-failed-retrieval.test.ts
index 0992e87ce32..a2c5c790379 100644
--- a/knowledge-fs/packages/api/src/workflow-failed-retrieval.test.ts
+++ b/knowledge-fs/packages/api/src/workflow-failed-retrieval.test.ts
@@ -9,6 +9,7 @@ const SPACE_ID = "10000000-0000-4000-8000-000000000001";
 const EVENT_ID = "10000000-0000-4000-8000-000000000002";
 const FIRST_GRANT_ID = "10000000-0000-4000-8000-000000000003";
 const RETRY_GRANT_ID = "10000000-0000-4000-8000-000000000004";
+const RETRIEVAL_TRACE_ID = "10000000-0000-4000-8000-000000000005";
 
 function baseInput(capabilityGrantId = FIRST_GRANT_ID) {
   return {
@@ -101,3 +102,49 @@ describe("workflow failed-retrieval capture", () => {
     },
   );
 });
+
+describe("workflow failed-retrieval history linkage", () => {
+  it("attaches the capture to the retrieval-test trace the node reports instead of a second record", async () => {
+    const { answerTraces, createBadCase, failedQueries, service } = setup("retrieval-miss");
+    const retrievalTrace = await createAnswerTraceRecorder({
+      now: () => "2026-08-12T00:00:00.000Z",
+      repository: answerTraces,
+    }).record({
+      capabilityGrantId: FIRST_GRANT_ID,
+      knowledgeSpaceId: SPACE_ID,
+      mode: "deep",
+      query: "发票号码在哪里？",
+      source: "workflow",
+      steps: [{ metadata: { resultCount: 0 }, name: "retrieval.test", status: "ok" }],
+      tenantId: "tenant-1",
+      traceId: RETRIEVAL_TRACE_ID,
+    });
+
+    const result = await service.capture({ ...baseInput(), retrievalTraceId: retrievalTrace.id });
+
+    expect(result.verdict).toBe("retrieval-miss");
+    expect(await answerTraces.getById(EVENT_ID)).toBeNull();
+    const failedQuery = await failedQueries.get({
+      candidateGrants: ["tenant:tenant-1"],
+      id: EVENT_ID,
+      knowledgeSpaceId: SPACE_ID,
+      subjectId: "dify-app:workflow-app",
+      tenantId: "tenant-1",
+    });
+    expect(failedQuery?.answerTraceId).toBe(RETRIEVAL_TRACE_ID);
+    expect(createBadCase).toHaveBeenCalledWith(
+      expect.objectContaining({ id: EVENT_ID, traceId: RETRIEVAL_TRACE_ID }),
+    );
+  });
+
+  it("falls back to a workflow-sourced trace keyed by the event when only a transport id is known", async () => {
+    const { answerTraces, service } = setup("coverage-gap");
+
+    await service.capture(baseInput());
+
+    expect(await answerTraces.getById(EVENT_ID)).toMatchObject({
+      id: EVENT_ID,
+      source: "workflow",
+    });
+  });
+});
diff --git a/knowledge-fs/packages/api/src/workflow-failed-retrieval.ts b/knowledge-fs/packages/api/src/workflow-failed-retrieval.ts
index e8e6d29c229..75eda5cd0ee 100644
--- a/knowledge-fs/packages/api/src/workflow-failed-retrieval.ts
+++ b/knowledge-fs/packages/api/src/workflow-failed-retrieval.ts
@@ -77,17 +77,16 @@ export function createWorkflowFailedRetrievalCaptureService({
       };
       let failedQuery = await failedQueries.get(lookup);
       let traceCapabilityGrantId = input.capabilityGrantId;
+      let answerTraceId = failedQuery?.answerTraceId ?? input.eventId;
 
       if (!failedQuery) {
-        traceCapabilityGrantId = await ensureWorkflowAnswerTrace(
-          answerTraces,
-          answerTraceRecorder,
-          input,
-        );
+        const trace = await ensureWorkflowAnswerTrace(answerTraces, answerTraceRecorder, input);
+        traceCapabilityGrantId = trace.capabilityGrantId;
+        answerTraceId = trace.id;
       }
       failedQuery = await failedQueries.captureWorkflowFailedRetrieval({
         actorSubjectId: input.actorSubjectId,
-        answerTraceId: input.eventId,
+        answerTraceId,
         candidateGrants: input.candidateGrants,
         capabilityGrantId: input.capabilityGrantId,
         id: input.eventId,
@@ -141,18 +140,36 @@ export function createWorkflowFailedRetrievalCaptureService({
           "Workflow retrieval returned no evidence even though the knowledge base appears to contain relevant answer material.",
         tags: ["workflow", "auto-captured", "retrieval-miss"],
         tenantId: input.tenantId,
-        traceId: input.eventId,
+        traceId: answerTraceId,
       });
       return { badCaseId: badCase.id, failedQueryId: failedQuery.id, verdict };
     },
   };
 }
 
+/**
+ * Resolve the AnswerTrace a failed workflow retrieval belongs to.
+ *
+ * The retrieval-tests route already records one trace per workflow run and returns its id as
+ * `answerTraceId`, which the node forwards as `retrievalTraceId`; attaching the failed query to
+ * that trace keeps one history record per retrieval and lets the overview classify the query as
+ * "no evidence". Older nodes send an opaque transport trace id instead, in which case a minimal
+ * trace keyed by the capture event id is created (or replayed) as before.
+ */
 async function ensureWorkflowAnswerTrace(
   answerTraces: Pick<AnswerTraceRepository, "get">,
   recorder: AnswerTraceRecorder,
   input: CaptureWorkflowFailedRetrievalInput,
-): Promise<string> {
+): Promise<{ readonly capabilityGrantId: string; readonly id: string }> {
+  if (UUID_PATTERN.test(input.retrievalTraceId) && input.retrievalTraceId !== input.eventId) {
+    const retrievalTrace = await answerTraces.get({
+      id: input.retrievalTraceId,
+      knowledgeSpaceId: input.knowledgeSpaceId,
+    });
+    if (retrievalTrace?.capabilityGrantId && retrievalTrace.query === input.query) {
+      return { capabilityGrantId: retrievalTrace.capabilityGrantId, id: retrievalTrace.id };
+    }
+  }
   const existing = await answerTraces.get({
     id: input.eventId,
     knowledgeSpaceId: input.knowledgeSpaceId,
@@ -162,13 +179,14 @@ async function ensureWorkflowAnswerTrace(
     if (!existing.capabilityGrantId) {
       throw new WorkflowFailedRetrievalReplayConflictError(input.eventId);
     }
-    return existing.capabilityGrantId;
+    return { capabilityGrantId: existing.capabilityGrantId, id: existing.id };
   }
   const recorded = await recorder.record({
     capabilityGrantId: input.capabilityGrantId,
     knowledgeSpaceId: input.knowledgeSpaceId,
     mode: input.mode,
     query: input.query,
+    source: "workflow",
     steps: [
       {
         metadata: {
@@ -188,9 +206,11 @@ async function ensureWorkflowAnswerTrace(
   if (!recorded.capabilityGrantId) {
     throw new Error("Workflow failed-retrieval AnswerTrace lost Capability provenance");
   }
-  return recorded.capabilityGrantId;
+  return { capabilityGrantId: recorded.capabilityGrantId, id: recorded.id };
 }
 
+const UUID_PATTERN = /^[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}$/iu;
+
 function assertWorkflowAnswerTraceReplay(
   existing: AnswerTrace,
   input: CaptureWorkflowFailedRetrievalInput,
diff --git a/knowledge-fs/packages/core/src/models.ts b/knowledge-fs/packages/core/src/models.ts
index 692adfadc31..dca0dfff470 100644
--- a/knowledge-fs/packages/core/src/models.ts
+++ b/knowledge-fs/packages/core/src/models.ts
@@ -1533,6 +1533,25 @@ export const AnswerTraceStepSchema = z.object({
 });
 export type AnswerTraceStep = z.infer<typeof AnswerTraceStepSchema>;
 
+/**
+ * Which caller produced an AnswerTrace. Console retrieval tests, KnowledgeFS workflow nodes,
+ * service API clients, agents, and MCP clients all record traces into the same history, and the
+ * console distinguishes them by this value. Rows written before the column existed are
+ * `retrieval_test`.
+ */
+export const AnswerTraceSourceSchema = z.enum([
+  "retrieval_test",
+  "workflow",
+  "service_api",
+  "agent",
+  "mcp",
+]);
+export type AnswerTraceSource = z.infer<typeof AnswerTraceSourceSchema>;
+
+export function answerTraceSourceOf(trace: { readonly source?: AnswerTraceSource | undefined }) {
+  return trace.source ?? "retrieval_test";
+}
+
 export const AnswerTraceSchema = z.object({
   /** Durable Capability v2 provenance; mutually exclusive with legacy member snapshots. */
   capabilityGrantId: UuidSchema.optional(),
@@ -1550,6 +1569,8 @@ export const AnswerTraceSchema = z.object({
     .optional(),
   query: z.string(),
   queryImages: z.array(QueryImageMetadataSchema).max(4).optional(),
+  /** Absent on legacy rows; read it through `answerTraceSourceOf`. */
+  source: AnswerTraceSourceSchema.optional(),
   /** Authenticated creator. Legacy traces without it are intentionally unreadable via the API. */
   subjectId: z.string().min(1).max(255).optional(),
   steps: z.array(AnswerTraceStepSchema),
diff --git a/knowledge-fs/packages/database/migrations/0053_answer_trace_source.postgres.sql b/knowledge-fs/packages/database/migrations/0053_answer_trace_source.postgres.sql
new file mode 100644
index 00000000000..2dbe9c035ef
--- /dev/null
+++ b/knowledge-fs/packages/database/migrations/0053_answer_trace_source.postgres.sql
@@ -0,0 +1,11 @@
+-- Knowledge Platform schema migration
+-- Migration id: 0053_answer_trace_source
+-- Dialect: postgres
+-- Answer traces record which caller produced them (console retrieval test, workflow node, service
+-- API, agent, MCP) so retrieval history can distinguish and filter them. NULL means retrieval_test.
+
+ALTER TABLE "answer_traces"
+  ADD COLUMN IF NOT EXISTS "source" TEXT;
+
+CREATE INDEX IF NOT EXISTS "answer_traces_space_source_created_idx"
+  ON "answer_traces" ("knowledge_space_id", "source", "created_at", "id");
diff --git a/knowledge-fs/packages/database/migrations/0053_answer_trace_source.tidb.sql b/knowledge-fs/packages/database/migrations/0053_answer_trace_source.tidb.sql
new file mode 100644
index 00000000000..62c7dda6889
--- /dev/null
+++ b/knowledge-fs/packages/database/migrations/0053_answer_trace_source.tidb.sql
@@ -0,0 +1,11 @@
+-- Knowledge Platform schema migration
+-- Migration id: 0053_answer_trace_source
+-- Dialect: tidb
+-- Answer traces record which caller produced them (console retrieval test, workflow node, service
+-- API, agent, MCP) so retrieval history can distinguish and filter them. NULL means retrieval_test.
+
+ALTER TABLE `answer_traces`
+  ADD COLUMN IF NOT EXISTS `source` VARCHAR(32) NULL;
+
+CREATE INDEX IF NOT EXISTS `answer_traces_space_source_created_idx`
+  ON `answer_traces` (`knowledge_space_id`, `source`, `created_at`, `id`);
diff --git a/knowledge-fs/packages/database/src/migration-artifacts.generated.ts b/knowledge-fs/packages/database/src/migration-artifacts.generated.ts
index 1a22941538f..98171cc0bf8 100644
--- a/knowledge-fs/packages/database/src/migration-artifacts.generated.ts
+++ b/knowledge-fs/packages/database/src/migration-artifacts.generated.ts
@@ -107,4 +107,6 @@ export const migrationArtifacts = [
   { content: "-- Knowledge Platform schema migration\n-- Migration id: 0051_failed_query_trace_outcome_index\n-- Dialect: tidb\n-- Supports Overview outcome joins by exact trace and trigger without scanning space history.\n\nCREATE INDEX IF NOT EXISTS `failed_queries_trace_trigger_idx`\n  ON `failed_queries` (\n    `tenant_id`, `knowledge_space_id`, `answer_trace_id`\n  );\n", path: "packages/database/migrations/0051_failed_query_trace_outcome_index.tidb.sql" },
   { content: "-- Knowledge Platform schema migration\n-- Migration id: 0052_namespace_source_previews\n-- Dialect: postgres\n\nCREATE TABLE IF NOT EXISTS \"namespace_source_preview_jobs\" (\n  \"id\" uuid PRIMARY KEY,\n  \"tenant_id\" varchar(255) NOT NULL,\n  \"account_id\" varchar(255) NOT NULL,\n  \"status\" varchar(32) NOT NULL,\n  \"provider_config\" jsonb NOT NULL,\n  \"configuration_fingerprint\" varchar(128) NOT NULL,\n  \"expires_at\" timestamptz NOT NULL,\n  \"consumed_at\" timestamptz NULL,\n  \"content_cleaned_at\" timestamptz NULL,\n  \"import_workflow_id\" uuid NULL,\n  \"error_code\" varchar(128) NULL,\n  \"created_at\" timestamptz NOT NULL,\n  \"updated_at\" timestamptz NOT NULL\n);\nCREATE INDEX IF NOT EXISTS \"namespace_source_preview_jobs_claim_idx\"\n  ON \"namespace_source_preview_jobs\" (\"status\", \"expires_at\", \"created_at\");\nCREATE INDEX IF NOT EXISTS \"namespace_source_preview_jobs_owner_idx\"\n  ON \"namespace_source_preview_jobs\" (\"tenant_id\", \"account_id\", \"created_at\");\nCREATE INDEX IF NOT EXISTS \"namespace_source_preview_jobs_cleanup_idx\"\n  ON \"namespace_source_preview_jobs\" (\"status\", \"content_cleaned_at\", \"updated_at\");\n\nCREATE TABLE IF NOT EXISTS \"namespace_source_preview_pages\" (\n  \"job_id\" uuid NOT NULL REFERENCES \"namespace_source_preview_jobs\" (\"id\") ON DELETE CASCADE,\n  \"page_id\" varchar(128) NOT NULL,\n  \"source_url\" varchar(4096) NOT NULL,\n  \"title\" varchar(500) NULL,\n  \"description\" text NULL,\n  \"content_hash\" varchar(64) NOT NULL,\n  \"content_object_key\" varchar(2048) NOT NULL,\n  \"created_at\" timestamptz NOT NULL,\n  PRIMARY KEY (\"job_id\", \"page_id\")\n);\n", path: "packages/database/migrations/0052_namespace_source_previews.postgres.sql" },
   { content: "-- Knowledge Platform schema migration\n-- Migration id: 0052_namespace_source_previews\n-- Dialect: tidb\n\nCREATE TABLE IF NOT EXISTS `namespace_source_preview_jobs` (\n  `id` char(36) PRIMARY KEY,\n  `tenant_id` varchar(255) NOT NULL,\n  `account_id` varchar(255) NOT NULL,\n  `status` varchar(32) NOT NULL,\n  `provider_config` json NOT NULL,\n  `configuration_fingerprint` varchar(128) NOT NULL,\n  `expires_at` datetime(6) NOT NULL,\n  `consumed_at` datetime(6) NULL,\n  `content_cleaned_at` datetime(6) NULL,\n  `import_workflow_id` char(36) NULL,\n  `error_code` varchar(128) NULL,\n  `created_at` datetime(6) NOT NULL,\n  `updated_at` datetime(6) NOT NULL,\n  INDEX `namespace_source_preview_jobs_claim_idx` (`status`, `expires_at`, `created_at`),\n  INDEX `namespace_source_preview_jobs_owner_idx` (`tenant_id`, `account_id`, `created_at`),\n  INDEX `namespace_source_preview_jobs_cleanup_idx` (`status`, `content_cleaned_at`, `updated_at`)\n);\n\nCREATE TABLE IF NOT EXISTS `namespace_source_preview_pages` (\n  `job_id` char(36) NOT NULL,\n  `page_id` varchar(128) NOT NULL,\n  `source_url` varchar(4096) NOT NULL,\n  `title` varchar(500) NULL,\n  `description` text NULL,\n  `content_hash` varchar(64) NOT NULL,\n  `content_object_key` varchar(2048) NOT NULL,\n  `created_at` datetime(6) NOT NULL,\n  PRIMARY KEY (`job_id`, `page_id`),\n  CONSTRAINT `namespace_source_preview_pages_job_fk` FOREIGN KEY (`job_id`) REFERENCES `namespace_source_preview_jobs` (`id`) ON DELETE CASCADE\n);\n", path: "packages/database/migrations/0052_namespace_source_previews.tidb.sql" },
+  { content: "-- Knowledge Platform schema migration\n-- Migration id: 0053_answer_trace_source\n-- Dialect: postgres\n-- Answer traces record which caller produced them (console retrieval test, workflow node, service\n-- API, agent, MCP) so retrieval history can distinguish and filter them. NULL means retrieval_test.\n\nALTER TABLE \"answer_traces\"\n  ADD COLUMN IF NOT EXISTS \"source\" TEXT;\n\nCREATE INDEX IF NOT EXISTS \"answer_traces_space_source_created_idx\"\n  ON \"answer_traces\" (\"knowledge_space_id\", \"source\", \"created_at\", \"id\");\n", path: "packages/database/migrations/0053_answer_trace_source.postgres.sql" },
+  { content: "-- Knowledge Platform schema migration\n-- Migration id: 0053_answer_trace_source\n-- Dialect: tidb\n-- Answer traces record which caller produced them (console retrieval test, workflow node, service\n-- API, agent, MCP) so retrieval history can distinguish and filter them. NULL means retrieval_test.\n\nALTER TABLE `answer_traces`\n  ADD COLUMN IF NOT EXISTS `source` VARCHAR(32) NULL;\n\nCREATE INDEX IF NOT EXISTS `answer_traces_space_source_created_idx`\n  ON `answer_traces` (`knowledge_space_id`, `source`, `created_at`, `id`);\n", path: "packages/database/migrations/0053_answer_trace_source.tidb.sql" },
 ] as const satisfies readonly MigrationArtifact[];
diff --git a/knowledge-fs/packages/database/src/migration-file.test.ts b/knowledge-fs/packages/database/src/migration-file.test.ts
index 787d153e2b2..fe81966824f 100644
--- a/knowledge-fs/packages/database/src/migration-file.test.ts
+++ b/knowledge-fs/packages/database/src/migration-file.test.ts
@@ -164,6 +164,8 @@ describe("migration file rendering", () => {
       "packages/database/migrations/0051_failed_query_trace_outcome_index.tidb.sql",
       "packages/database/migrations/0052_namespace_source_previews.postgres.sql",
       "packages/database/migrations/0052_namespace_source_previews.tidb.sql",
+      "packages/database/migrations/0053_answer_trace_source.postgres.sql",
+      "packages/database/migrations/0053_answer_trace_source.tidb.sql",
     ]);
     const workflowCapturePostgres = artifacts.find(
       (artifact) =>
@@ -946,6 +948,7 @@ describe("migration file rendering", () => {
       "packages/database/migrations/0050_deletion_job_capability_provenance.postgres.sql",
       "packages/database/migrations/0051_failed_query_trace_outcome_index.postgres.sql",
       "packages/database/migrations/0052_namespace_source_previews.postgres.sql",
+      "packages/database/migrations/0053_answer_trace_source.postgres.sql",
     ]);
     expect(
       getPendingMigrationArtifacts({
@@ -1002,6 +1005,7 @@ describe("migration file rendering", () => {
           "0050_deletion_job_capability_provenance",
           "0051_failed_query_trace_outcome_index",
           "0052_namespace_source_previews",
+          "0053_answer_trace_source",
         ],
         dialect: "postgres",
       }),
diff --git a/packages/contracts/generated/api/console/knowledge-fs/types.gen.ts b/packages/contracts/generated/api/console/knowledge-fs/types.gen.ts
index bf56808be57..eb304054823 100644
--- a/packages/contracts/generated/api/console/knowledge-fs/types.gen.ts
+++ b/packages/contracts/generated/api/console/knowledge-fs/types.gen.ts
@@ -1197,6 +1197,7 @@ export type KnowledgeFsAnswerTraceResponse = {
   mode: 'auto' | 'deep' | 'fast' | 'research'
   query: string
   query_images?: Array<KnowledgeFsQueryImageResponse>
+  source?: KnowledgeFsTraceSource
   steps: Array<KnowledgeFsAnswerTraceStepResponse>
 }
 
@@ -2164,9 +2165,12 @@ export type KnowledgeFsTraceResponse = {
   query_images?: Array<KnowledgeFsQueryImageResponse>
   result_count: number
   scores: KnowledgeFsTraceScoresResponse
+  source?: KnowledgeFsTraceSource
   stages: Array<KnowledgeFsTraceStageResponse>
 }
 
+export type KnowledgeFsTraceSource = 'agent' | 'mcp' | 'retrieval_test' | 'service_api' | 'workflow'
+
 export type KnowledgeFsAnswerTraceStepResponse = {
   ended_at?: string | null
   metadata: {
@@ -4404,6 +4408,7 @@ export type GetKnowledgeFsSpacesByControlSpaceIdTracesData = {
   }
   query?: {
     cursor?: string
+    source?: 'agent' | 'mcp' | 'retrieval_test' | 'service_api' | 'workflow'
   }
   url: '/knowledge-fs/spaces/{control_space_id}/traces'
 }
diff --git a/packages/contracts/generated/api/console/knowledge-fs/zod.gen.ts b/packages/contracts/generated/api/console/knowledge-fs/zod.gen.ts
index 4c233ce3215..abf0eb381e1 100644
--- a/packages/contracts/generated/api/console/knowledge-fs/zod.gen.ts
+++ b/packages/contracts/generated/api/console/knowledge-fs/zod.gen.ts
@@ -2159,6 +2159,14 @@ export const zKnowledgeFsSpaceListResponse = z.object({
   page: z.int(),
 })
 
+export const zKnowledgeFsTraceSource = z.enum([
+  'agent',
+  'mcp',
+  'retrieval_test',
+  'service_api',
+  'workflow',
+])
+
 /**
  * KnowledgeFSAnswerTraceStepResponse
  */
@@ -2181,6 +2189,7 @@ export const zKnowledgeFsAnswerTraceResponse = z.object({
   mode: z.enum(['auto', 'deep', 'fast', 'research']),
   query: z.string(),
   query_images: z.array(zKnowledgeFsQueryImageResponse).optional(),
+  source: zKnowledgeFsTraceSource.optional().default('retrieval_test'),
   steps: z.array(zKnowledgeFsAnswerTraceStepResponse),
 })
 
@@ -3060,6 +3069,7 @@ export const zKnowledgeFsTraceResponse = z.object({
   query_images: z.array(zKnowledgeFsQueryImageResponse).optional(),
   result_count: z.int().gte(0),
   scores: zKnowledgeFsTraceScoresResponse,
+  source: zKnowledgeFsTraceSource.optional().default('retrieval_test'),
   stages: z.array(zKnowledgeFsTraceStageResponse),
 })
 
@@ -4651,6 +4661,7 @@ export const zGetKnowledgeFsSpacesByControlSpaceIdTracesPath = z.object({
 
 export const zGetKnowledgeFsSpacesByControlSpaceIdTracesQuery = z.object({
   cursor: z.string().min(1).max(1000).optional(),
+  source: z.enum(['agent', 'mcp', 'retrieval_test', 'service_api', 'workflow']).optional(),
 })
 
 /**
diff --git a/packages/contracts/generated/api/service/types.gen.ts b/packages/contracts/generated/api/service/types.gen.ts
index 336a9fd1839..0b9cf1a129e 100644
--- a/packages/contracts/generated/api/service/types.gen.ts
+++ b/packages/contracts/generated/api/service/types.gen.ts
@@ -1269,6 +1269,7 @@ export type KnowledgeFsAnswerTraceResponse = {
   mode: 'auto' | 'deep' | 'fast' | 'research'
   query: string
   query_images?: Array<KnowledgeFsQueryImageResponse>
+  source?: KnowledgeFsTraceSource
   steps: Array<KnowledgeFsAnswerTraceStepResponse>
 }
 
@@ -2487,6 +2488,7 @@ export type KnowledgeFsTraceResponse = {
   query_images?: Array<KnowledgeFsQueryImageResponse>
   result_count: number
   scores: KnowledgeFsTraceScoresResponse
+  source?: KnowledgeFsTraceSource
   stages: Array<KnowledgeFsTraceStageResponse>
 }
 
@@ -2496,6 +2498,8 @@ export type KnowledgeFsTraceScoresResponse = {
   retrieval?: number | null
 }
 
+export type KnowledgeFsTraceSource = 'agent' | 'mcp' | 'retrieval_test' | 'service_api' | 'workflow'
+
 export type KnowledgeFsTraceStageResponse = {
   candidate_count?: number | null
   name: string
@@ -6048,6 +6052,7 @@ export type GetKnowledgeFsSpacesByControlSpaceIdTracesData = {
   }
   query?: {
     cursor?: string
+    source?: 'agent' | 'mcp' | 'retrieval_test' | 'service_api' | 'workflow'
   }
   url: '/knowledge-fs/spaces/{control_space_id}/traces'
 }
diff --git a/packages/contracts/generated/api/service/zod.gen.ts b/packages/contracts/generated/api/service/zod.gen.ts
index 3a22750bd40..36a74ca8560 100644
--- a/packages/contracts/generated/api/service/zod.gen.ts
+++ b/packages/contracts/generated/api/service/zod.gen.ts
@@ -2047,20 +2047,6 @@ export const zKnowledgeFsQueryImageResponse = z.object({
   upload_file_id: z.string(),
 })
 
-/**
- * KnowledgeFSAnswerTraceResponse
- */
-export const zKnowledgeFsAnswerTraceResponse = z.object({
-  created_at: z.iso.datetime(),
-  evidence_bundle_id: z.string().nullish(),
-  id: z.string(),
-  knowledge_space_id: z.string(),
-  mode: z.enum(['auto', 'deep', 'fast', 'research']),
-  query: z.string(),
-  query_images: z.array(zKnowledgeFsQueryImageResponse).optional(),
-  steps: z.array(zKnowledgeFsAnswerTraceStepResponse),
-})
-
 /**
  * KnowledgeFSReadinessCapabilities
  */
@@ -2905,6 +2891,29 @@ export const zKnowledgeFsTraceScoresResponse = z.object({
   retrieval: z.number().gte(0).lte(1).nullish(),
 })
 
+export const zKnowledgeFsTraceSource = z.enum([
+  'agent',
+  'mcp',
+  'retrieval_test',
+  'service_api',
+  'workflow',
+])
+
+/**
+ * KnowledgeFSAnswerTraceResponse
+ */
+export const zKnowledgeFsAnswerTraceResponse = z.object({
+  created_at: z.iso.datetime(),
+  evidence_bundle_id: z.string().nullish(),
+  id: z.string(),
+  knowledge_space_id: z.string(),
+  mode: z.enum(['auto', 'deep', 'fast', 'research']),
+  query: z.string(),
+  query_images: z.array(zKnowledgeFsQueryImageResponse).optional(),
+  source: zKnowledgeFsTraceSource.optional().default('retrieval_test'),
+  steps: z.array(zKnowledgeFsAnswerTraceStepResponse),
+})
+
 /**
  * KnowledgeFSTraceStageResponse
  */
@@ -2932,6 +2941,7 @@ export const zKnowledgeFsTraceResponse = z.object({
   query_images: z.array(zKnowledgeFsQueryImageResponse).optional(),
   result_count: z.int().gte(0),
   scores: zKnowledgeFsTraceScoresResponse,
+  source: zKnowledgeFsTraceSource.optional().default('retrieval_test'),
   stages: z.array(zKnowledgeFsTraceStageResponse),
 })
 
@@ -5958,6 +5968,7 @@ export const zGetKnowledgeFsSpacesByControlSpaceIdTracesPath = z.object({
 
 export const zGetKnowledgeFsSpacesByControlSpaceIdTracesQuery = z.object({
   cursor: z.string().min(1).max(1000).optional(),
+  source: z.enum(['agent', 'mcp', 'retrieval_test', 'service_api', 'workflow']).optional(),
 })
 
 /**
diff --git a/web/features/new-rag/retrieval/__tests__/model.spec.ts b/web/features/new-rag/retrieval/__tests__/model.spec.ts
index 2732d808901..e8d2b117661 100644
--- a/web/features/new-rag/retrieval/__tests__/model.spec.ts
+++ b/web/features/new-rag/retrieval/__tests__/model.spec.ts
@@ -264,3 +264,35 @@ describe('retrieval test model', () => {
     ).toBe(false)
   })
 })
+
+describe('retrieval record sources', () => {
+  it('keeps the trace source and defaults legacy traces to retrieval tests', () => {
+    const [workflow, legacy] = retrievalTestRecords(
+      [
+        {
+          completed: true,
+          created_at: '2026-09-04T10:00:00.000Z',
+          id: 'trace-workflow',
+          mode: 'fast',
+          query: 'From a workflow node',
+          result_count: 3,
+          source: 'workflow',
+        } as KnowledgeFsTraceResponse,
+        {
+          completed: true,
+          created_at: '2026-09-04T09:00:00.000Z',
+          id: 'trace-legacy',
+          mode: 'fast',
+          query: 'From the console',
+          result_count: 1,
+        } as KnowledgeFsTraceResponse,
+      ],
+      [],
+    )
+
+    expect(workflow).toEqual(expect.objectContaining({ id: 'trace-workflow', source: 'workflow' }))
+    expect(legacy).toEqual(
+      expect.objectContaining({ id: 'trace-legacy', source: 'retrieval_test' }),
+    )
+  })
+})
diff --git a/web/features/new-rag/retrieval/history-panel.tsx b/web/features/new-rag/retrieval/history-panel.tsx
index b0f983833c0..b02cca63efe 100644
--- a/web/features/new-rag/retrieval/history-panel.tsx
+++ b/web/features/new-rag/retrieval/history-panel.tsx
@@ -1,6 +1,8 @@
 'use client'
 
+import type { RetrievalHistorySourceFilter } from './state/scoped'
 import { Button } from '@langgenius/dify-ui/button'
+import { SegmentedControl, SegmentedControlItem } from '@langgenius/dify-ui/segmented-control'
 import { useAtomValue, useSetAtom } from 'jotai'
 import { useTranslation } from 'react-i18next'
 import { RecordButton } from './history'
@@ -9,20 +11,40 @@ import {
   retrievalHistoryFactsAtom,
   selectRetrievalRecordAtom,
 } from './state/graph'
+import { retrievalHistorySourceFilterAtom } from './state/scoped'
+
+const sourceFilters = ['all', 'retrieval_test', 'workflow'] as const
 
 export function RetrievalHistoryPanel() {
   const { t } = useTranslation('knowledgeSpace')
-  const { activeRecordKey, displayRecords, hasNextPage, isFetchingNextPage } =
+  const { activeRecordKey, displayRecords, hasNextPage, isFetchingNextPage, sourceFilter } =
     useAtomValue(retrievalHistoryFactsAtom)
   const selectRecord = useSetAtom(selectRetrievalRecordAtom)
   const loadMore = useSetAtom(loadMoreRetrievalHistoryAtom)
+  const setSourceFilter = useSetAtom(retrievalHistorySourceFilterAtom)
 
   return (
     <div className="mt-3 flex min-h-0 flex-1 flex-col pt-6">
-      <div className="flex shrink-0 items-center pb-2 pl-3">
+      <div className="flex shrink-0 items-center justify-between gap-2 pr-3 pb-2 pl-3">
         <h2 className="system-xs-medium text-text-tertiary">
           {t(($) => $['retrievalTest.records'])}
         </h2>
+        <SegmentedControl<RetrievalHistorySourceFilter>
+          aria-label={t(($) => $['retrievalTest.sourceFilter.label'])}
+          className="shrink-0"
+          value={sourceFilter}
+          onValueChange={setSourceFilter}
+        >
+          {sourceFilters.map((filter) => (
+            <SegmentedControlItem<RetrievalHistorySourceFilter>
+              key={filter}
+              value={filter}
+              className="px-1.5 py-0.5 system-2xs-medium"
+            >
+              {t(($) => $[`retrievalTest.sourceFilter.${filter}`])}
+            </SegmentedControlItem>
+          ))}
+        </SegmentedControl>
       </div>
       <div className="min-h-0 flex-1 scrollbar-none overflow-y-auto">
         {displayRecords.length > 0 ? (
diff --git a/web/features/new-rag/retrieval/history.tsx b/web/features/new-rag/retrieval/history.tsx
index cc06160ae10..e8accf5fa55 100644
--- a/web/features/new-rag/retrieval/history.tsx
+++ b/web/features/new-rag/retrieval/history.tsx
@@ -575,6 +575,14 @@ export function RecordButton({
               {queryImageCount}
             </span>
           )}
+          {record.kind === 'trace' && record.source !== 'retrieval_test' && (
+            <span
+              className="inline-flex shrink-0 items-center rounded-[5px] border border-divider-deep px-1 system-2xs-medium-uppercase text-text-tertiary"
+              data-testid="retrieval-record-source"
+            >
+              {t(($) => $[`retrievalTest.source.${record.source}`])}
+            </span>
+          )}
         </span>
         <span className="mt-1.5 flex items-center gap-1 system-xs-regular text-text-tertiary">
           <span className={cn('min-w-0 flex-1 truncate', failed && 'text-text-destructive')}>
diff --git a/web/features/new-rag/retrieval/model.ts b/web/features/new-rag/retrieval/model.ts
index 26146f2deed..33d920e33b9 100644
--- a/web/features/new-rag/retrieval/model.ts
+++ b/web/features/new-rag/retrieval/model.ts
@@ -2,8 +2,16 @@ import type {
   KnowledgeFsQueryImageResponse,
   KnowledgeFsResearchTaskResponse,
   KnowledgeFsTraceResponse,
+  KnowledgeFsTraceSource,
 } from '@dify/contracts/api/console/knowledge-fs/types.gen'
 
+/** Who produced a persisted trace: the console retrieval test or another caller. */
+export type RetrievalTestSource = KnowledgeFsTraceSource
+
+export function retrievalTestSource(source?: KnowledgeFsTraceSource | null): RetrievalTestSource {
+  return source ?? 'retrieval_test'
+}
+
 export type RetrievalTestMode = 'deep' | 'fast' | 'research'
 
 export function normalizedRetrievalTestMode(mode?: string): RetrievalTestMode {
@@ -74,6 +82,8 @@ export type RetrievalTestRecord =
       query: string
       queryImages?: RetrievalQueryImage[]
       resultCount?: number
+      /** Console retrieval test, workflow node, service API, agent, or MCP. */
+      source: RetrievalTestSource
       status: 'completed' | 'failed'
     }
   | {
@@ -398,6 +408,7 @@ export function retrievalTestRecords(
         ...(trace.result_count !== null && trace.result_count !== undefined
           ? { resultCount: trace.result_count }
           : {}),
+        source: retrievalTestSource(trace.source),
         status: trace.completed ? 'completed' : 'failed',
       })),
     ...researchTasks.map((task): RetrievalTestRecord => ({
diff --git a/web/features/new-rag/retrieval/state/graph.ts b/web/features/new-rag/retrieval/state/graph.ts
index 733f2761738..36a4fadc47c 100644
--- a/web/features/new-rag/retrieval/state/graph.ts
+++ b/web/features/new-rag/retrieval/state/graph.ts
@@ -24,6 +24,7 @@ import {
   retrievalAdmittedResearchTasksAtom,
   retrievalComposerDraftAtom,
   retrievalComposerImagesAtom,
+  retrievalHistorySourceFilterAtom,
   retrievalLocalRunAtom,
   retrievalLocalSelectedAtom,
   retrievalRecordImagesAtom,
@@ -38,10 +39,17 @@ const tracesQueryAtom = atomWithInfiniteQuery((get) =>
   consoleQuery.knowledgeFs.spaces.byControlSpaceId.traces.get.infiniteOptions({
     getNextPageParam: (lastPage) => lastPage.next_cursor ?? undefined,
     initialPageParam: null as string | null,
-    input: (pageParam) => ({
-      params: { control_space_id: get(retrievalKnowledgeSpaceIdAtom) },
-      ...(typeof pageParam === 'string' ? { query: { cursor: pageParam } } : {}),
-    }),
+    input: (pageParam) => {
+      const sourceFilter = get(retrievalHistorySourceFilterAtom)
+      const query = {
+        ...(typeof pageParam === 'string' ? { cursor: pageParam } : {}),
+        ...(sourceFilter === 'all' ? {} : { source: sourceFilter }),
+      }
+      return {
+        params: { control_space_id: get(retrievalKnowledgeSpaceIdAtom) },
+        ...(Object.keys(query).length > 0 ? { query } : {}),
+      }
+    },
     refetchInterval: get(retrievalLocalRunAtom)?.status === 'running' ? 1000 : false,
   }),
 )
@@ -119,7 +127,14 @@ const retrievalDisplayRecordsAtom = atom((get) => {
     localRun?.traceId &&
     records.some((record) => record.kind === 'trace' && record.id === localRun.traceId),
   )
-  return localRecord && !traceAlreadyListed ? [localRecord, ...records] : records
+  const displayRecords = localRecord && !traceAlreadyListed ? [localRecord, ...records] : records
+  // The server already filters persisted traces by source; local runs and research tasks are
+  // console retrieval tests, so they only belong to the "all" and "retrieval_test" views.
+  const sourceFilter = get(retrievalHistorySourceFilterAtom)
+  if (sourceFilter === 'all') return displayRecords
+  return displayRecords.filter((record) =>
+    record.kind === 'trace' ? record.source === sourceFilter : sourceFilter === 'retrieval_test',
+  )
 })
 
 const requestedSelectionAtom = atom((get) => {
@@ -440,6 +455,7 @@ export const retrievalHistoryFactsAtom = atom((get) => {
     displayRecords,
     hasNextPage: Boolean(tracesQuery.hasNextPage || researchTasksQuery.hasNextPage),
     isFetchingNextPage: tracesQuery.isFetchingNextPage || researchTasksQuery.isFetchingNextPage,
+    sourceFilter: get(retrievalHistorySourceFilterAtom),
   }
 })
 
diff --git a/web/features/new-rag/retrieval/state/scoped.ts b/web/features/new-rag/retrieval/state/scoped.ts
index 63c07663d91..e8e6b2fd35f 100644
--- a/web/features/new-rag/retrieval/state/scoped.ts
+++ b/web/features/new-rag/retrieval/state/scoped.ts
@@ -62,6 +62,9 @@ export const retrievalAdmittedResearchTasksAtom = atom<
   Record<string, KnowledgeFsResearchTaskResponse>
 >({})
 export const retrievalResearchRetryPendingAtom = atom(false)
+/** History filter: every record, only console retrieval tests, or only workflow-node queries. */
+export type RetrievalHistorySourceFilter = 'all' | 'retrieval_test' | 'workflow'
+export const retrievalHistorySourceFilterAtom = atom<RetrievalHistorySourceFilter>('all')
 
 export const retrievalScopedAtoms = [
   retrievalComposerDraftAtom,
@@ -73,5 +76,6 @@ export const retrievalScopedAtoms = [
   retrievalResearchEventsAtom,
   retrievalAdmittedResearchTasksAtom,
   retrievalResearchRetryPendingAtom,
+  retrievalHistorySourceFilterAtom,
   retrievalRuntimeBridgeAtom,
 ] as const
diff --git a/web/i18n/ar-TN/knowledge-space.json b/web/i18n/ar-TN/knowledge-space.json
index 4958c5733b0..2d7a3ee8aa3 100644
--- a/web/i18n/ar-TN/knowledge-space.json
+++ b/web/i18n/ar-TN/knowledge-space.json
@@ -506,6 +506,15 @@
   "retrievalTest.savedGoldenQuestion": "تم الحفظ كسؤال مرجعي",
   "retrievalTest.score": "الدرجة",
   "retrievalTest.showAllChunks": "عرض جميع المقاطع وعددها {{count}}",
+  "retrievalTest.source.agent": "الوكيل",
+  "retrievalTest.source.mcp": "MCP",
+  "retrievalTest.source.retrieval_test": "اختبار الاسترجاع",
+  "retrievalTest.source.service_api": "Service API",
+  "retrievalTest.source.workflow": "سير العمل",
+  "retrievalTest.sourceFilter.all": "الكل",
+  "retrievalTest.sourceFilter.label": "مصدر السجل",
+  "retrievalTest.sourceFilter.retrieval_test": "اختبارات الاسترجاع",
+  "retrievalTest.sourceFilter.workflow": "سير العمل",
   "retrievalTest.startResearch": "بدء البحث",
   "retrievalTest.title": "اختبار الاسترجاع",
   "retrievalTest.viewInQuality": "عرض في الجودة",
diff --git a/web/i18n/de-DE/knowledge-space.json b/web/i18n/de-DE/knowledge-space.json
index 8dee3cd5547..c4201f3cd31 100644
--- a/web/i18n/de-DE/knowledge-space.json
+++ b/web/i18n/de-DE/knowledge-space.json
@@ -506,6 +506,15 @@
   "retrievalTest.savedGoldenQuestion": "Als Referenzfrage gespeichert",
   "retrievalTest.score": "Punktzahl",
   "retrievalTest.showAllChunks": "Alle {{count}} Abschnitte anzeigen",
+  "retrievalTest.source.agent": "Agent",
+  "retrievalTest.source.mcp": "MCP",
+  "retrievalTest.source.retrieval_test": "Abruftest",
+  "retrievalTest.source.service_api": "Service-API",
+  "retrievalTest.source.workflow": "Workflow",
+  "retrievalTest.sourceFilter.all": "Alle",
+  "retrievalTest.sourceFilter.label": "Quelle des Eintrags",
+  "retrievalTest.sourceFilter.retrieval_test": "Abruftests",
+  "retrievalTest.sourceFilter.workflow": "Workflow",
   "retrievalTest.startResearch": "Recherche starten",
   "retrievalTest.title": "Abruf testen",
   "retrievalTest.viewInQuality": "In „Qualität“ anzeigen",
diff --git a/web/i18n/en-US/knowledge-space.json b/web/i18n/en-US/knowledge-space.json
index c1b5036d812..09c12568478 100644
--- a/web/i18n/en-US/knowledge-space.json
+++ b/web/i18n/en-US/knowledge-space.json
@@ -506,6 +506,15 @@
   "retrievalTest.savedGoldenQuestion": "Saved as golden question",
   "retrievalTest.score": "Score",
   "retrievalTest.showAllChunks": "Show all {{count}} chunks",
+  "retrievalTest.source.agent": "Agent",
+  "retrievalTest.source.mcp": "MCP",
+  "retrievalTest.source.retrieval_test": "Retrieval test",
+  "retrievalTest.source.service_api": "Service API",
+  "retrievalTest.source.workflow": "Workflow",
+  "retrievalTest.sourceFilter.all": "All",
+  "retrievalTest.sourceFilter.label": "Record source",
+  "retrievalTest.sourceFilter.retrieval_test": "Retrieval tests",
+  "retrievalTest.sourceFilter.workflow": "Workflow",
   "retrievalTest.startResearch": "Start research",
   "retrievalTest.title": "Retrieval Test",
   "retrievalTest.viewInQuality": "View in Quality",
diff --git a/web/i18n/es-ES/knowledge-space.json b/web/i18n/es-ES/knowledge-space.json
index 91ac0f5187e..4ab1d24eae8 100644
--- a/web/i18n/es-ES/knowledge-space.json
+++ b/web/i18n/es-ES/knowledge-space.json
@@ -506,6 +506,15 @@
   "retrievalTest.savedGoldenQuestion": "Guardado como pregunta de referencia",
   "retrievalTest.score": "Puntuación",
   "retrievalTest.showAllChunks": "Mostrar los {{count}} fragmentos",
+  "retrievalTest.source.agent": "Agente",
+  "retrievalTest.source.mcp": "MCP",
+  "retrievalTest.source.retrieval_test": "Prueba de recuperación",
+  "retrievalTest.source.service_api": "API de servicio",
+  "retrievalTest.source.workflow": "Flujo de trabajo",
+  "retrievalTest.sourceFilter.all": "Todos",
+  "retrievalTest.sourceFilter.label": "Origen del registro",
+  "retrievalTest.sourceFilter.retrieval_test": "Pruebas de recuperación",
+  "retrievalTest.sourceFilter.workflow": "Flujo de trabajo",
   "retrievalTest.startResearch": "Iniciar investigación",
   "retrievalTest.title": "Prueba de recuperación",
   "retrievalTest.viewInQuality": "Ver en Calidad",
diff --git a/web/i18n/fa-IR/knowledge-space.json b/web/i18n/fa-IR/knowledge-space.json
index 2850da8b262..eae4302f6a3 100644
--- a/web/i18n/fa-IR/knowledge-space.json
+++ b/web/i18n/fa-IR/knowledge-space.json
@@ -506,6 +506,15 @@
   "retrievalTest.savedGoldenQuestion": "به‌عنوان پرسش مرجع ذخیره شد",
   "retrievalTest.score": "امتیاز",
   "retrievalTest.showAllChunks": "نمایش همه {{count}} قطعه",
+  "retrievalTest.source.agent": "عامل",
+  "retrievalTest.source.mcp": "MCP",
+  "retrievalTest.source.retrieval_test": "آزمون بازیابی",
+  "retrievalTest.source.service_api": "Service API",
+  "retrievalTest.source.workflow": "گردش کار",
+  "retrievalTest.sourceFilter.all": "همه",
+  "retrievalTest.sourceFilter.label": "منبع رکورد",
+  "retrievalTest.sourceFilter.retrieval_test": "آزمون‌های بازیابی",
+  "retrievalTest.sourceFilter.workflow": "گردش کار",
   "retrievalTest.startResearch": "شروع پژوهش",
   "retrievalTest.title": "آزمون بازیابی",
   "retrievalTest.viewInQuality": "نمایش در بخش کیفیت",
diff --git a/web/i18n/fr-FR/knowledge-space.json b/web/i18n/fr-FR/knowledge-space.json
index 03d9536e304..5152ffc461a 100644
--- a/web/i18n/fr-FR/knowledge-space.json
+++ b/web/i18n/fr-FR/knowledge-space.json
@@ -506,6 +506,15 @@
   "retrievalTest.savedGoldenQuestion": "Enregistré comme question de référence",
   "retrievalTest.score": "Score",
   "retrievalTest.showAllChunks": "Afficher les {{count}} segments",
+  "retrievalTest.source.agent": "Agent",
+  "retrievalTest.source.mcp": "MCP",
+  "retrievalTest.source.retrieval_test": "Test de récupération",
+  "retrievalTest.source.service_api": "API de service",
+  "retrievalTest.source.workflow": "Workflow",
+  "retrievalTest.sourceFilter.all": "Tout",
+  "retrievalTest.sourceFilter.label": "Origine de l'enregistrement",
+  "retrievalTest.sourceFilter.retrieval_test": "Tests de récupération",
+  "retrievalTest.sourceFilter.workflow": "Workflow",
   "retrievalTest.startResearch": "Démarrer la recherche",
   "retrievalTest.title": "Test de récupération",
   "retrievalTest.viewInQuality": "Afficher dans Qualité",
diff --git a/web/i18n/hi-IN/knowledge-space.json b/web/i18n/hi-IN/knowledge-space.json
index 2735870c699..14eb99c9e32 100644
--- a/web/i18n/hi-IN/knowledge-space.json
+++ b/web/i18n/hi-IN/knowledge-space.json
@@ -506,6 +506,15 @@
   "retrievalTest.savedGoldenQuestion": "संदर्भ प्रश्न के रूप में सहेजा गया",
   "retrievalTest.score": "स्कोर",
   "retrievalTest.showAllChunks": "सभी {{count}} खंड दिखाएँ",
+  "retrievalTest.source.agent": "एजेंट",
+  "retrievalTest.source.mcp": "MCP",
+  "retrievalTest.source.retrieval_test": "पुनर्प्राप्ति परीक्षण",
+  "retrievalTest.source.service_api": "Service API",
+  "retrievalTest.source.workflow": "वर्कफ़्लो",
+  "retrievalTest.sourceFilter.all": "सभी",
+  "retrievalTest.sourceFilter.label": "रिकॉर्ड का स्रोत",
+  "retrievalTest.sourceFilter.retrieval_test": "पुनर्प्राप्ति परीक्षण",
+  "retrievalTest.sourceFilter.workflow": "वर्कफ़्लो",
   "retrievalTest.startResearch": "रिसर्च शुरू करें",
   "retrievalTest.title": "रिट्रीवल परीक्षण",
   "retrievalTest.viewInQuality": "गुणवत्ता में देखें",
diff --git a/web/i18n/id-ID/knowledge-space.json b/web/i18n/id-ID/knowledge-space.json
index 02d9537eddd..117a015d7cf 100644
--- a/web/i18n/id-ID/knowledge-space.json
+++ b/web/i18n/id-ID/knowledge-space.json
@@ -506,6 +506,15 @@
   "retrievalTest.savedGoldenQuestion": "Disimpan sebagai pertanyaan referensi",
   "retrievalTest.score": "Skor",
   "retrievalTest.showAllChunks": "Tampilkan semua {{count}} potongan",
+  "retrievalTest.source.agent": "Agen",
+  "retrievalTest.source.mcp": "MCP",
+  "retrievalTest.source.retrieval_test": "Uji pengambilan",
+  "retrievalTest.source.service_api": "Service API",
+  "retrievalTest.source.workflow": "Alur kerja",
+  "retrievalTest.sourceFilter.all": "Semua",
+  "retrievalTest.sourceFilter.label": "Sumber catatan",
+  "retrievalTest.sourceFilter.retrieval_test": "Uji pengambilan",
+  "retrievalTest.sourceFilter.workflow": "Alur kerja",
   "retrievalTest.startResearch": "Mulai riset",
   "retrievalTest.title": "Uji Pengambilan",
   "retrievalTest.viewInQuality": "Lihat di Kualitas",
diff --git a/web/i18n/it-IT/knowledge-space.json b/web/i18n/it-IT/knowledge-space.json
index a3792bfcf3f..614c896c69e 100644
--- a/web/i18n/it-IT/knowledge-space.json
+++ b/web/i18n/it-IT/knowledge-space.json
@@ -506,6 +506,15 @@
   "retrievalTest.savedGoldenQuestion": "Salvato come domanda di riferimento",
   "retrievalTest.score": "Punteggio",
   "retrievalTest.showAllChunks": "Mostra tutti i {{count}} segmenti",
+  "retrievalTest.source.agent": "Agente",
+  "retrievalTest.source.mcp": "MCP",
+  "retrievalTest.source.retrieval_test": "Test di recupero",
+  "retrievalTest.source.service_api": "API di servizio",
+  "retrievalTest.source.workflow": "Workflow",
+  "retrievalTest.sourceFilter.all": "Tutti",
+  "retrievalTest.sourceFilter.label": "Origine del record",
+  "retrievalTest.sourceFilter.retrieval_test": "Test di recupero",
+  "retrievalTest.sourceFilter.workflow": "Workflow",
   "retrievalTest.startResearch": "Avvia ricerca",
   "retrievalTest.title": "Test di recupero",
   "retrievalTest.viewInQuality": "Visualizza in Qualità",
diff --git a/web/i18n/ja-JP/knowledge-space.json b/web/i18n/ja-JP/knowledge-space.json
index 4d1ede010d8..6775833d83b 100644
--- a/web/i18n/ja-JP/knowledge-space.json
+++ b/web/i18n/ja-JP/knowledge-space.json
@@ -506,6 +506,15 @@
   "retrievalTest.savedGoldenQuestion": "参考質問として保存しました",
   "retrievalTest.score": "スコア",
   "retrievalTest.showAllChunks": "全 {{count}} チャンクを表示",
+  "retrievalTest.source.agent": "エージェント",
+  "retrievalTest.source.mcp": "MCP",
+  "retrievalTest.source.retrieval_test": "検索テスト",
+  "retrievalTest.source.service_api": "Service API",
+  "retrievalTest.source.workflow": "ワークフロー",
+  "retrievalTest.sourceFilter.all": "すべて",
+  "retrievalTest.sourceFilter.label": "記録の発生元",
+  "retrievalTest.sourceFilter.retrieval_test": "検索テスト",
+  "retrievalTest.sourceFilter.workflow": "ワークフロー",
   "retrievalTest.startResearch": "調査を開始",
   "retrievalTest.title": "検索テスト",
   "retrievalTest.viewInQuality": "品質で表示",
diff --git a/web/i18n/ko-KR/knowledge-space.json b/web/i18n/ko-KR/knowledge-space.json
index 8d8e120e7d6..a27c498d48e 100644
--- a/web/i18n/ko-KR/knowledge-space.json
+++ b/web/i18n/ko-KR/knowledge-space.json
@@ -506,6 +506,15 @@
   "retrievalTest.savedGoldenQuestion": "참고 질문으로 저장됨",
   "retrievalTest.score": "점수",
   "retrievalTest.showAllChunks": "청크 {{count}}개 모두 표시",
+  "retrievalTest.source.agent": "에이전트",
+  "retrievalTest.source.mcp": "MCP",
+  "retrievalTest.source.retrieval_test": "검색 테스트",
+  "retrievalTest.source.service_api": "Service API",
+  "retrievalTest.source.workflow": "워크플로",
+  "retrievalTest.sourceFilter.all": "전체",
+  "retrievalTest.sourceFilter.label": "기록 출처",
+  "retrievalTest.sourceFilter.retrieval_test": "검색 테스트",
+  "retrievalTest.sourceFilter.workflow": "워크플로",
   "retrievalTest.startResearch": "리서치 시작",
   "retrievalTest.title": "검색 테스트",
   "retrievalTest.viewInQuality": "품질에서 보기",
diff --git a/web/i18n/lo-LA/knowledge-space.json b/web/i18n/lo-LA/knowledge-space.json
index 6dd81071f7a..5a1f939df92 100644
--- a/web/i18n/lo-LA/knowledge-space.json
+++ b/web/i18n/lo-LA/knowledge-space.json
@@ -506,6 +506,15 @@
   "retrievalTest.savedGoldenQuestion": "ບັນທຶກເປັນຄຳຖາມມາດຕະຖານແລ້ວ",
   "retrievalTest.score": "ຄະແນນ",
   "retrievalTest.showAllChunks": "ສະແດງທັງໝົດ {{count}} ສ່ວນ",
+  "retrievalTest.source.agent": "ຕົວແທນ",
+  "retrievalTest.source.mcp": "MCP",
+  "retrievalTest.source.retrieval_test": "ທົດສອບການດຶງຂໍ້ມູນ",
+  "retrievalTest.source.service_api": "Service API",
+  "retrievalTest.source.workflow": "ຂັ້ນຕອນການເຮັດວຽກ",
+  "retrievalTest.sourceFilter.all": "ທັງໝົດ",
+  "retrievalTest.sourceFilter.label": "ແຫຼ່ງທີ່ມາຂອງບັນທຶກ",
+  "retrievalTest.sourceFilter.retrieval_test": "ທົດສອບການດຶງຂໍ້ມູນ",
+  "retrievalTest.sourceFilter.workflow": "ຂັ້ນຕອນການເຮັດວຽກ",
   "retrievalTest.startResearch": "ເລີ່ມການຄົ້ນຄວ້າ",
   "retrievalTest.title": "ທົດສອບການຄົ້ນຄືນ",
   "retrievalTest.viewInQuality": "ເບິ່ງໃນໜ້າຄຸນະພາບ",
diff --git a/web/i18n/nl-NL/knowledge-space.json b/web/i18n/nl-NL/knowledge-space.json
index 3908f9483ab..370a3cbedc0 100644
--- a/web/i18n/nl-NL/knowledge-space.json
+++ b/web/i18n/nl-NL/knowledge-space.json
@@ -506,6 +506,15 @@
   "retrievalTest.savedGoldenQuestion": "Opgeslagen als referentievraag",
   "retrievalTest.score": "Score",
   "retrievalTest.showAllChunks": "Alle {{count}} fragmenten tonen",
+  "retrievalTest.source.agent": "Agent",
+  "retrievalTest.source.mcp": "MCP",
+  "retrievalTest.source.retrieval_test": "Ophaaltest",
+  "retrievalTest.source.service_api": "Service-API",
+  "retrievalTest.source.workflow": "Workflow",
+  "retrievalTest.sourceFilter.all": "Alle",
+  "retrievalTest.sourceFilter.label": "Bron van record",
+  "retrievalTest.sourceFilter.retrieval_test": "Ophaaltests",
+  "retrievalTest.sourceFilter.workflow": "Workflow",
   "retrievalTest.startResearch": "Onderzoek starten",
   "retrievalTest.title": "Ophalen testen",
   "retrievalTest.viewInQuality": "Bekijken in Kwaliteit",
diff --git a/web/i18n/pl-PL/knowledge-space.json b/web/i18n/pl-PL/knowledge-space.json
index 57934851ab3..3d0ec58e2f5 100644
--- a/web/i18n/pl-PL/knowledge-space.json
+++ b/web/i18n/pl-PL/knowledge-space.json
@@ -506,6 +506,15 @@
   "retrievalTest.savedGoldenQuestion": "Zapisano jako pytanie referencyjne",
   "retrievalTest.score": "Wynik",
   "retrievalTest.showAllChunks": "Pokaż wszystkie fragmenty ({{count}})",
+  "retrievalTest.source.agent": "Agent",
+  "retrievalTest.source.mcp": "MCP",
+  "retrievalTest.source.retrieval_test": "Test wyszukiwania",
+  "retrievalTest.source.service_api": "API usługi",
+  "retrievalTest.source.workflow": "Przepływ pracy",
+  "retrievalTest.sourceFilter.all": "Wszystkie",
+  "retrievalTest.sourceFilter.label": "Źródło rekordu",
+  "retrievalTest.sourceFilter.retrieval_test": "Testy wyszukiwania",
+  "retrievalTest.sourceFilter.workflow": "Przepływ pracy",
   "retrievalTest.startResearch": "Rozpocznij badanie",
   "retrievalTest.title": "Test wyszukiwania",
   "retrievalTest.viewInQuality": "Zobacz w sekcji Jakość",
diff --git a/web/i18n/pt-BR/knowledge-space.json b/web/i18n/pt-BR/knowledge-space.json
index f136464634e..6f6dd0e1b34 100644
--- a/web/i18n/pt-BR/knowledge-space.json
+++ b/web/i18n/pt-BR/knowledge-space.json
@@ -506,6 +506,15 @@
   "retrievalTest.savedGoldenQuestion": "Salvo como pergunta de referência",
   "retrievalTest.score": "Pontuação",
   "retrievalTest.showAllChunks": "Mostrar todos os {{count}} trechos",
+  "retrievalTest.source.agent": "Agente",
+  "retrievalTest.source.mcp": "MCP",
+  "retrievalTest.source.retrieval_test": "Teste de recuperação",
+  "retrievalTest.source.service_api": "API de serviço",
+  "retrievalTest.source.workflow": "Fluxo de trabalho",
+  "retrievalTest.sourceFilter.all": "Todos",
+  "retrievalTest.sourceFilter.label": "Origem do registro",
+  "retrievalTest.sourceFilter.retrieval_test": "Testes de recuperação",
+  "retrievalTest.sourceFilter.workflow": "Fluxo de trabalho",
   "retrievalTest.startResearch": "Iniciar pesquisa",
   "retrievalTest.title": "Teste de recuperação",
   "retrievalTest.viewInQuality": "Ver em Qualidade",
diff --git a/web/i18n/ro-RO/knowledge-space.json b/web/i18n/ro-RO/knowledge-space.json
index 993e96e1f6a..a825a43b452 100644
--- a/web/i18n/ro-RO/knowledge-space.json
+++ b/web/i18n/ro-RO/knowledge-space.json
@@ -506,6 +506,15 @@
   "retrievalTest.savedGoldenQuestion": "Salvat ca întrebare de referință",
   "retrievalTest.score": "Scor",
   "retrievalTest.showAllChunks": "Afișează toate cele {{count}} fragmente",
+  "retrievalTest.source.agent": "Agent",
+  "retrievalTest.source.mcp": "MCP",
+  "retrievalTest.source.retrieval_test": "Test de regăsire",
+  "retrievalTest.source.service_api": "API de serviciu",
+  "retrievalTest.source.workflow": "Flux de lucru",
+  "retrievalTest.sourceFilter.all": "Toate",
+  "retrievalTest.sourceFilter.label": "Sursa înregistrării",
+  "retrievalTest.sourceFilter.retrieval_test": "Teste de regăsire",
+  "retrievalTest.sourceFilter.workflow": "Flux de lucru",
   "retrievalTest.startResearch": "Începe cercetarea",
   "retrievalTest.title": "Test de recuperare",
   "retrievalTest.viewInQuality": "Vezi în Calitate",
diff --git a/web/i18n/ru-RU/knowledge-space.json b/web/i18n/ru-RU/knowledge-space.json
index 63354725f37..5c2b4c46539 100644
--- a/web/i18n/ru-RU/knowledge-space.json
+++ b/web/i18n/ru-RU/knowledge-space.json
@@ -506,6 +506,15 @@
   "retrievalTest.savedGoldenQuestion": "Сохранено как эталонный вопрос",
   "retrievalTest.score": "Оценка",
   "retrievalTest.showAllChunks": "Показать все фрагменты ({{count}})",
+  "retrievalTest.source.agent": "Агент",
+  "retrievalTest.source.mcp": "MCP",
+  "retrievalTest.source.retrieval_test": "Тест поиска",
+  "retrievalTest.source.service_api": "Service API",
+  "retrievalTest.source.workflow": "Рабочий процесс",
+  "retrievalTest.sourceFilter.all": "Все",
+  "retrievalTest.sourceFilter.label": "Источник записи",
+  "retrievalTest.sourceFilter.retrieval_test": "Тесты поиска",
+  "retrievalTest.sourceFilter.workflow": "Рабочий процесс",
   "retrievalTest.startResearch": "Начать исследование",
   "retrievalTest.title": "Проверка поиска",
   "retrievalTest.viewInQuality": "Открыть в разделе «Качество»",
diff --git a/web/i18n/sl-SI/knowledge-space.json b/web/i18n/sl-SI/knowledge-space.json
index f124a739cd8..028de78b0a4 100644
--- a/web/i18n/sl-SI/knowledge-space.json
+++ b/web/i18n/sl-SI/knowledge-space.json
@@ -506,6 +506,15 @@
   "retrievalTest.savedGoldenQuestion": "Shranjeno kot referenčno vprašanje",
   "retrievalTest.score": "Rezultat",
   "retrievalTest.showAllChunks": "Prikaži vseh {{count}} odsekov",
+  "retrievalTest.source.agent": "Agent",
+  "retrievalTest.source.mcp": "MCP",
+  "retrievalTest.source.retrieval_test": "Preizkus pridobivanja",
+  "retrievalTest.source.service_api": "Storitveni API",
+  "retrievalTest.source.workflow": "Potek dela",
+  "retrievalTest.sourceFilter.all": "Vse",
+  "retrievalTest.sourceFilter.label": "Vir zapisa",
+  "retrievalTest.sourceFilter.retrieval_test": "Preizkusi pridobivanja",
+  "retrievalTest.sourceFilter.workflow": "Potek dela",
   "retrievalTest.startResearch": "Začni raziskavo",
   "retrievalTest.title": "Preskus pridobivanja",
   "retrievalTest.viewInQuality": "Prikaži v razdelku Kakovost",
diff --git a/web/i18n/th-TH/knowledge-space.json b/web/i18n/th-TH/knowledge-space.json
index 63ad7b28be0..7866ad2e498 100644
--- a/web/i18n/th-TH/knowledge-space.json
+++ b/web/i18n/th-TH/knowledge-space.json
@@ -506,6 +506,15 @@
   "retrievalTest.savedGoldenQuestion": "บันทึกเป็นคำถามอ้างอิงแล้ว",
   "retrievalTest.score": "คะแนน",
   "retrievalTest.showAllChunks": "แสดงส่วนเนื้อหาทั้งหมด {{count}} รายการ",
+  "retrievalTest.source.agent": "เอเจนต์",
+  "retrievalTest.source.mcp": "MCP",
+  "retrievalTest.source.retrieval_test": "ทดสอบการค้นคืน",
+  "retrievalTest.source.service_api": "Service API",
+  "retrievalTest.source.workflow": "เวิร์กโฟลว์",
+  "retrievalTest.sourceFilter.all": "ทั้งหมด",
+  "retrievalTest.sourceFilter.label": "ที่มาของบันทึก",
+  "retrievalTest.sourceFilter.retrieval_test": "ทดสอบการค้นคืน",
+  "retrievalTest.sourceFilter.workflow": "เวิร์กโฟลว์",
   "retrievalTest.startResearch": "เริ่มการวิจัย",
   "retrievalTest.title": "ทดสอบการค้นคืน",
   "retrievalTest.viewInQuality": "ดูในคุณภาพ",
diff --git a/web/i18n/tr-TR/knowledge-space.json b/web/i18n/tr-TR/knowledge-space.json
index 6f6f0fd66a0..5336e7da5b5 100644
--- a/web/i18n/tr-TR/knowledge-space.json
+++ b/web/i18n/tr-TR/knowledge-space.json
@@ -506,6 +506,15 @@
   "retrievalTest.savedGoldenQuestion": "Referans sorusu olarak kaydedildi",
   "retrievalTest.score": "Puan",
   "retrievalTest.showAllChunks": "{{count}} parçanın tümünü göster",
+  "retrievalTest.source.agent": "Ajan",
+  "retrievalTest.source.mcp": "MCP",
+  "retrievalTest.source.retrieval_test": "Getirme testi",
+  "retrievalTest.source.service_api": "Servis API'si",
+  "retrievalTest.source.workflow": "İş akışı",
+  "retrievalTest.sourceFilter.all": "Tümü",
+  "retrievalTest.sourceFilter.label": "Kayıt kaynağı",
+  "retrievalTest.sourceFilter.retrieval_test": "Getirme testleri",
+  "retrievalTest.sourceFilter.workflow": "İş akışı",
   "retrievalTest.startResearch": "Araştırmayı başlat",
   "retrievalTest.title": "Getirme Testi",
   "retrievalTest.viewInQuality": "Kalite bölümünde görüntüle",
diff --git a/web/i18n/uk-UA/knowledge-space.json b/web/i18n/uk-UA/knowledge-space.json
index 197632c904c..75dcdcac529 100644
--- a/web/i18n/uk-UA/knowledge-space.json
+++ b/web/i18n/uk-UA/knowledge-space.json
@@ -506,6 +506,15 @@
   "retrievalTest.savedGoldenQuestion": "Збережено як еталонне запитання",
   "retrievalTest.score": "Оцінка",
   "retrievalTest.showAllChunks": "Показати всі фрагменти ({{count}})",
+  "retrievalTest.source.agent": "Агент",
+  "retrievalTest.source.mcp": "MCP",
+  "retrievalTest.source.retrieval_test": "Тест пошуку",
+  "retrievalTest.source.service_api": "Service API",
+  "retrievalTest.source.workflow": "Робочий процес",
+  "retrievalTest.sourceFilter.all": "Усі",
+  "retrievalTest.sourceFilter.label": "Джерело запису",
+  "retrievalTest.sourceFilter.retrieval_test": "Тести пошуку",
+  "retrievalTest.sourceFilter.workflow": "Робочий процес",
   "retrievalTest.startResearch": "Почати дослідження",
   "retrievalTest.title": "Перевірка пошуку",
   "retrievalTest.viewInQuality": "Відкрити в розділі «Якість»",
diff --git a/web/i18n/vi-VN/knowledge-space.json b/web/i18n/vi-VN/knowledge-space.json
index cecb29a7fc2..4015f7b5978 100644
--- a/web/i18n/vi-VN/knowledge-space.json
+++ b/web/i18n/vi-VN/knowledge-space.json
@@ -506,6 +506,15 @@
   "retrievalTest.savedGoldenQuestion": "Đã lưu làm câu hỏi tham chiếu",
   "retrievalTest.score": "Điểm",
   "retrievalTest.showAllChunks": "Hiển thị tất cả {{count}} đoạn",
+  "retrievalTest.source.agent": "Agent",
+  "retrievalTest.source.mcp": "MCP",
+  "retrievalTest.source.retrieval_test": "Kiểm tra truy xuất",
+  "retrievalTest.source.service_api": "Service API",
+  "retrievalTest.source.workflow": "Quy trình",
+  "retrievalTest.sourceFilter.all": "Tất cả",
+  "retrievalTest.sourceFilter.label": "Nguồn bản ghi",
+  "retrievalTest.sourceFilter.retrieval_test": "Kiểm tra truy xuất",
+  "retrievalTest.sourceFilter.workflow": "Quy trình",
   "retrievalTest.startResearch": "Bắt đầu nghiên cứu",
   "retrievalTest.title": "Kiểm thử truy xuất",
   "retrievalTest.viewInQuality": "Xem trong Chất lượng",
diff --git a/web/i18n/zh-Hans/knowledge-space.json b/web/i18n/zh-Hans/knowledge-space.json
index 53d9a048587..ab15b9a251f 100644
--- a/web/i18n/zh-Hans/knowledge-space.json
+++ b/web/i18n/zh-Hans/knowledge-space.json
@@ -506,6 +506,15 @@
   "retrievalTest.savedGoldenQuestion": "已保存为参考问题",
   "retrievalTest.score": "得分",
   "retrievalTest.showAllChunks": "显示全部 {{count}} 个分段",
+  "retrievalTest.source.agent": "Agent",
+  "retrievalTest.source.mcp": "MCP",
+  "retrievalTest.source.retrieval_test": "检索测试",
+  "retrievalTest.source.service_api": "Service API",
+  "retrievalTest.source.workflow": "工作流",
+  "retrievalTest.sourceFilter.all": "全部",
+  "retrievalTest.sourceFilter.label": "记录来源",
+  "retrievalTest.sourceFilter.retrieval_test": "检索测试",
+  "retrievalTest.sourceFilter.workflow": "工作流",
   "retrievalTest.startResearch": "开始研究",
   "retrievalTest.title": "检索测试",
   "retrievalTest.viewInQuality": "前往“质量”页查看",
diff --git a/web/i18n/zh-Hant/knowledge-space.json b/web/i18n/zh-Hant/knowledge-space.json
index 55fc18d14e4..4d048fb3571 100644
--- a/web/i18n/zh-Hant/knowledge-space.json
+++ b/web/i18n/zh-Hant/knowledge-space.json
@@ -506,6 +506,15 @@
   "retrievalTest.savedGoldenQuestion": "已儲存為參考問題",
   "retrievalTest.score": "分數",
   "retrievalTest.showAllChunks": "顯示全部 {{count}} 個片段",
+  "retrievalTest.source.agent": "Agent",
+  "retrievalTest.source.mcp": "MCP",
+  "retrievalTest.source.retrieval_test": "檢索測試",
+  "retrievalTest.source.service_api": "Service API",
+  "retrievalTest.source.workflow": "工作流",
+  "retrievalTest.sourceFilter.all": "全部",
+  "retrievalTest.sourceFilter.label": "記錄來源",
+  "retrievalTest.sourceFilter.retrieval_test": "檢索測試",
+  "retrievalTest.sourceFilter.workflow": "工作流",
   "retrievalTest.startResearch": "開始研究",
   "retrievalTest.title": "檢索測試",
   "retrievalTest.viewInQuality": "在品質中查看",