Avoid top-level composition in tool schemas

skills/product-data/schemas/export_dataset_records.input.schema.json

@@ -1,8 +1,14 @@
 {
   "type": "object",
   "properties": {
-    "records": {"type": "array"},
-    "records_path": {"type": "string"},
+    "records": {
+      "type": "array",
+      "description": "Canonical records. Provide exactly one of records or records_path."
+    },
+    "records_path": {
+      "type": "string",
+      "description": "Path to records JSON/JSONL. Provide exactly one of records or records_path."
+    },
     "output_path": {"type": "string"},
     "output_dir": {"type": "string"},
     "output_format": {
@@ -13,9 +19,5 @@
     "require_validation_ok": {"type": "boolean", "default": true},
     "overwrite": {"type": "boolean", "default": true},
     "export_table": {"type": "boolean", "default": true}
-  },
-  "anyOf": [
-    {"required": ["records"]},
-    {"required": ["records_path"]}
-  ]
+  }
 }

skills/product-data/schemas/export_dataset_table.input.schema.json

@@ -1,15 +1,17 @@
 {
   "type": "object",
   "properties": {
-    "records": {"type": "array"},
-    "records_path": {"type": "string"},
+    "records": {
+      "type": "array",
+      "description": "Canonical records. Provide exactly one of records or records_path."
+    },
+    "records_path": {
+      "type": "string",
+      "description": "Path to records JSON/JSONL. Provide exactly one of records or records_path."
+    },
     "output_path": {"type": "string"},
     "output_dir": {"type": "string"},
     "require_validation_ok": {"type": "boolean", "default": true},
     "overwrite": {"type": "boolean", "default": true}
-  },
-  "anyOf": [
-    {"required": ["records"]},
-    {"required": ["records_path"]}
-  ]
+  }
 }

skills/product-data/schemas/normalize_dataset_draft.input.schema.json

@@ -1,8 +1,14 @@
 {
   "type": "object",
   "properties": {
-    "draft_text": {"type": "string"},
-    "draft_path": {"type": "string"},
+    "draft_text": {
+      "type": "string",
+      "description": "Dataset draft text. Provide exactly one of draft_text or draft_path."
+    },
+    "draft_path": {
+      "type": "string",
+      "description": "Path to UTF-8 dataset draft text. Provide exactly one of draft_text or draft_path."
+    },
     "batch_id": {"type": "string", "default": "aabbccdd"},
     "source_type": {
       "type": "string",
@@ -12,9 +18,5 @@
     "base_timestamp": {"type": "integer"},
     "timestamp_step_ms": {"type": "integer", "minimum": 1, "default": 60000},
     "default_request_id": {"type": "string", "default": "aabbccdd"}
-  },
-  "anyOf": [
-    {"required": ["draft_text"]},
-    {"required": ["draft_path"]}
-  ]
+  }
 }

skills/product-data/schemas/validate_dataset_records.input.schema.json

@@ -1,11 +1,13 @@
 {
   "type": "object",
   "properties": {
-    "records": {"type": "array"},
-    "records_path": {"type": "string"}
+    "records": {
+      "type": "array",
+      "description": "Canonical records. Provide exactly one of records or records_path."
     },
-  "anyOf": [
-    {"required": ["records"]},
-    {"required": ["records_path"]}
-  ]
+    "records_path": {
+      "type": "string",
+      "description": "Path to records JSON/JSONL. Provide exactly one of records or records_path."
+    }
+  }
 }

src/agent_tool_specs/data_agent.py

@@ -437,11 +437,11 @@ def build_data_agent_tools(handlers: Mapping[str, ToolHandler]) -> list[AgentToo
                 'properties': {
                     'draft_text': {
                         'type': 'string',
-                        'description': 'Dataset draft text in dataset draft text v1 format.',
+                        'description': 'Dataset draft text in dataset draft text v1 format. Provide exactly one of draft_text or draft_path.',
                     },
                     'draft_path': {
                         'type': 'string',
-                        'description': 'Path to a UTF-8 dataset draft text file. Prefer this for large drafts.',
+                        'description': 'Path to a UTF-8 dataset draft text file. Prefer this for large drafts. Provide exactly one of draft_text or draft_path.',
                     },
                     'batch_id': {
                         'type': 'string',
@@ -469,10 +469,6 @@ def build_data_agent_tools(handlers: Mapping[str, ToolHandler]) -> list[AgentToo
                         'description': 'Required for generated data; returned by data_agent_confirm_generation_plan.',
                     },
                 },
-                'anyOf': [
-                    {'required': ['draft_text']},
-                    {'required': ['draft_path']},
-                ],
             },
             handler=resolve_handler(handlers, 'data_agent_normalize_dataset_draft', 'data-agent'),
         ),
@@ -490,17 +486,13 @@ def build_data_agent_tools(handlers: Mapping[str, ToolHandler]) -> list[AgentToo
                             {'type': 'array'},
                             {'type': 'string'},
                         ],
-                        'description': 'Canonical records as an array, or a JSON string containing the array.',
+                        'description': 'Canonical records as an array, or a JSON string containing the array. Provide exactly one of records or records_path.',
                     },
                     'records_path': {
                         'type': 'string',
-                        'description': 'Path to records JSON/JSONL. Prefer this when records are large.',
+                        'description': 'Path to records JSON/JSONL. Prefer this when records are large. Provide exactly one of records or records_path.',
                     },
                 },
-                'anyOf': [
-                    {'required': ['records']},
-                    {'required': ['records_path']},
-                ],
             },
             handler=resolve_handler(handlers, 'data_agent_validate_dataset_records', 'data-agent'),
         ),
@@ -519,11 +511,11 @@ def build_data_agent_tools(handlers: Mapping[str, ToolHandler]) -> list[AgentToo
                             {'type': 'array'},
                             {'type': 'string'},
                         ],
-                        'description': 'Canonical records as an array, or a JSON string containing the array.',
+                        'description': 'Canonical records as an array, or a JSON string containing the array. Provide exactly one of records or records_path.',
                     },
                     'records_path': {
                         'type': 'string',
-                        'description': 'Path to records JSON/JSONL. Prefer this when records are large.',
+                        'description': 'Path to records JSON/JSONL. Prefer this when records are large. Provide exactly one of records or records_path.',
                     },
                     'output_path': {
                         'type': 'string',
@@ -552,10 +544,6 @@ def build_data_agent_tools(handlers: Mapping[str, ToolHandler]) -> list[AgentToo
                     },
                 },
                 'required': ['output_path'],
-                'anyOf': [
-                    {'required': ['records']},
-                    {'required': ['records_path']},
-                ],
             },
             handler=resolve_handler(handlers, 'data_agent_export_dataset_records', 'data-agent'),
         ),

tests/test_data_agent_records.py

@@ -647,6 +647,17 @@ target: Agent(tag='地图导航')
         self.assertEqual(export_payload['record_count'], 1)
         self.assertTrue(table_exists)
 
+    def test_tool_schemas_avoid_top_level_composition_keywords(self) -> None:
+        # 部分 Bedrock/Anthropic 兼容后端不接受顶层 anyOf/oneOf/allOf。
+        blocked_keywords = {'anyOf', 'oneOf', 'allOf'}
+        violations = [
+            name
+            for name, tool in default_tool_registry().items()
+            if blocked_keywords.intersection(tool.parameters)
+        ]
+
+        self.assertEqual(violations, [])
+
     def test_data_agent_tools_route_relative_outputs_to_session_output(self) -> None:
         with tempfile.TemporaryDirectory() as tmp_dir:
             root = Path(tmp_dir)