Fix prompt formatting biases affecting JSON output

src/crewai/tools/base_tool.py

@@ -135,15 +135,25 @@ def _set_args_schema(self):
             )
 
     def _generate_description(self):
-        args_schema = {
-            name: {
-                "description": field.description,
-                "type": BaseTool._get_arg_annotations(field.annotation),
+        import json
+        import logging
+
+        logger = logging.getLogger(__name__)
+
+        try:
+            args_schema = {
+                name: {
+                    "description": field.description,
+                    "type": BaseTool._get_arg_annotations(field.annotation),
+                }
+                for name, field in self.args_schema.model_fields.items()
             }
-            for name, field in self.args_schema.model_fields.items()
-        }
-
-        self.description = f"Tool Name: {self.name}\nTool Arguments: {args_schema}\nTool Description: {self.description}"
+            args_json = json.dumps(args_schema)
+        except Exception as e:
+            logger.warning(f"Failed to serialize args schema: {e}")
+            args_json = str(args_schema)
+
+        self.description = f"Tool Name: {self.name}\nTool Arguments: {args_json}\nTool Description: {self.description}"
 
     @staticmethod
     def _get_arg_annotations(annotation: type[Any] | None) -> str:

src/crewai/translations/en.json

@@ -9,7 +9,7 @@
     "task": "\nCurrent Task: {input}\n\nBegin! This is VERY important to you, use the tools available and give your best Final Answer, your job depends on it!\n\nThought:",
     "memory": "\n\n# Useful context: \n{memory}",
     "role_playing": "You are {role}. {backstory}\nYour personal goal is: {goal}",
-    "tools": "\nYou ONLY have access to the following tools, and should NEVER make up tools that are not listed here:\n\n{tools}\n\nUse the following format:\n\nThought: you should always think about what to do\nAction: the action to take, only one name of [{tool_names}], just the name, exactly as it's written.\nAction Input: the input to the action, just a simple python dictionary, enclosed in curly braces, using \" to wrap keys and values.\nObservation: the result of the action\n\nOnce all necessary information is gathered:\n\nThought: I now know the final answer\nFinal Answer: the final answer to the original input question\n",
+    "tools": "\nYou ONLY have access to the following tools, and should NEVER make up tools that are not listed here:\n\n{tools}\n\nUse the following format:\n\nThought: you should always think about what to do\nAction: the action to take, only one name of [{tool_names}], just the name, exactly as it's written.\nAction Input: the input to the action, just a simple python dictionary, enclosed in curly braces, using double quotes (\") to wrap keys and values.\nObservation: the result of the action\n\nOnce all necessary information is gathered:\n\nThought: I now know the final answer\nFinal Answer: the final answer to the original input question\n",
     "no_tools": "\nTo give my best complete final answer to the task use the exact following format:\n\nThought: I now can give a great answer\nFinal Answer: Your final answer must be the great and the most complete as possible, it must be outcome described.\n\nI MUST use these formats, my job depends on it!",
     "format": "I MUST either use a tool (use one at time) OR give my best final answer not both at the same time. To Use the following format:\n\nThought: you should always think about what to do\nAction: the action to take, should be one of [{tool_names}]\nAction Input: the input to the action, dictionary enclosed in curly braces\nObservation: the result of the action\n... (this Thought/Action/Action Input/Result can repeat N times)\nThought: I now can give a great answer\nFinal Answer: Your final answer must be the great and the most complete as possible, it must be outcome described\n\n",
     "final_answer_format": "If you don't need to use any more tools, you must give your best complete final answer, make sure it satisfies the expected criteria, use the EXACT format below:\n\nThought: I now can give a great answer\nFinal Answer: my best complete final answer to the task.\n\n",

tests/tools/test_json_edge_cases.py

@@ -0,0 +1,85 @@
+import json
+from unittest.mock import MagicMock
+
+import pytest
+from pydantic import BaseModel, Field
+
+from crewai.tools import BaseTool
+from crewai.tools.tool_usage import ToolUsage
+
+
+class TestComplexInput(BaseModel):
+    special_chars: str = Field(
+        ..., description="Parameter with special characters: \"'\\{}[]"
+    )
+    nested_dict: dict = Field(
+        ..., description="A nested dictionary parameter"
+    )
+    unicode_text: str = Field(
+        ..., description="Text with unicode characters: 你好, こんにちは, مرحبا"
+    )
+
+
+class TestComplexTool(BaseTool):
+    name: str = "Complex JSON Tool"
+    description: str = "A tool for testing complex JSON formatting"
+    args_schema: type[BaseModel] = TestComplexInput
+
+    def _run(self, special_chars: str, nested_dict: dict, unicode_text: str) -> str:
+        return f"Processed complex input successfully"
+
+
+def test_complex_json_formatting():
+    """Test that complex JSON with special characters and nested structures is formatted correctly."""
+    tool = TestComplexTool()
+
+    assert "Tool Arguments:" in tool.description
+
+    description_parts = tool.description.split("Tool Arguments: ")
+    json_str = description_parts[1].split("\nTool Description:")[0]
+
+    parsed_json = json.loads(json_str)
+
+    assert "special_chars" in parsed_json
+    assert "nested_dict" in parsed_json
+    assert "unicode_text" in parsed_json
+
+    assert "\"'\\{}[]" in parsed_json["special_chars"]["description"]
+
+    assert "你好" in parsed_json["unicode_text"]["description"]
+    assert "こんにちは" in parsed_json["unicode_text"]["description"]
+    assert "مرحبا" in parsed_json["unicode_text"]["description"]
+
+
+def test_complex_tool_usage_render():
+    """Test that complex tool usage renders with proper JSON formatting."""
+    tool = TestComplexTool()
+
+    tool_usage = ToolUsage(
+        tools_handler=MagicMock(),
+        tools=[tool],
+        original_tools=[tool],
+        tools_description="Tool for testing complex JSON formatting",
+        tools_names="test_complex_tool",
+        task=MagicMock(),
+        function_calling_llm=MagicMock(),
+        agent=MagicMock(),
+        action=MagicMock(),
+    )
+
+    rendered = tool_usage._render()
+
+    rendered_parts = rendered.split("Tool Arguments: ")
+    if len(rendered_parts) > 1:
+        json_str = rendered_parts[1].split("\nTool Description:")[0]
+
+        try:
+            parsed_json = json.loads(json_str)
+            assert True  # If we get here, JSON parsing succeeded
+
+            assert "special_chars" in parsed_json
+            assert "nested_dict" in parsed_json
+            assert "unicode_text" in parsed_json
+
+        except json.JSONDecodeError:
+            assert False, "The rendered tool arguments are not valid JSON"

tests/tools/test_json_formatting.py

@@ -0,0 +1,78 @@
+import json
+from unittest.mock import MagicMock
+
+import pytest
+from pydantic import BaseModel, Field
+
+from crewai.tools import BaseTool
+from crewai.tools.tool_usage import ToolUsage
+
+
+class TestJsonInput(BaseModel):
+    test_param: str = Field(
+        ..., description="A test parameter"
+    )
+    another_param: int = Field(
+        ..., description="Another test parameter"
+    )
+
+
+class TestJsonTool(BaseTool):
+    name: str = "Test JSON Tool"
+    description: str = "A tool for testing JSON formatting"
+    args_schema: type[BaseModel] = TestJsonInput
+
+    def _run(self, test_param: str, another_param: int) -> str:
+        return f"Received {test_param} and {another_param}"
+
+
+def test_tool_description_json_formatting():
+    """Test that the tool description uses proper JSON formatting with double quotes."""
+    tool = TestJsonTool()
+
+    assert "Tool Arguments:" in tool.description
+
+    description_parts = tool.description.split("Tool Arguments: ")
+    json_str = description_parts[1].split("\nTool Description:")[0]
+
+    parsed_json = json.loads(json_str)
+
+    assert "test_param" in parsed_json
+    assert "another_param" in parsed_json
+
+    assert '"test_param"' in json_str
+    assert '"another_param"' in json_str
+    assert "'" not in json_str  # No single quotes should be present
+
+
+def test_tool_usage_json_formatting():
+    """Test that the tool usage renders with proper JSON formatting."""
+    tool = TestJsonTool()
+
+    tool_usage = ToolUsage(
+        tools_handler=MagicMock(),
+        tools=[tool],
+        original_tools=[tool],
+        tools_description="Tool for testing JSON formatting",
+        tools_names="test_json_tool",
+        task=MagicMock(),
+        function_calling_llm=MagicMock(),
+        agent=MagicMock(),
+        action=MagicMock(),
+    )
+
+    rendered = tool_usage._render()
+
+    rendered_parts = rendered.split("Tool Arguments: ")
+    if len(rendered_parts) > 1:
+        json_str = rendered_parts[1].split("\nTool Description:")[0]
+
+        try:
+            parsed_json = json.loads(json_str)
+            assert True  # If we get here, JSON parsing succeeded
+        except json.JSONDecodeError:
+            assert False, "The rendered tool arguments are not valid JSON"
+
+        assert '"test_param"' in json_str
+        assert '"another_param"' in json_str
+        assert "'" not in json_str  # No single quotes should be present

tests/tools/test_tool_usage.py

@@ -102,22 +102,22 @@ def test_tool_usage_render():
 
     rendered = tool_usage._render()
 
-    # Updated checks to match the actual output
+    # Updated checks to match the actual output with JSON formatting
     assert "Tool Name: Random Number Generator" in rendered
     assert "Tool Arguments:" in rendered
     assert (
-        "'min_value': {'description': 'The minimum value of the range (inclusive)', 'type': 'int'}"
+        '"min_value": {"description": "The minimum value of the range (inclusive)", "type": "int"}'
         in rendered
     )
     assert (
-        "'max_value': {'description': 'The maximum value of the range (inclusive)', 'type': 'int'}"
+        '"max_value": {"description": "The maximum value of the range (inclusive)", "type": "int"}'
         in rendered
     )
     assert (
         "Tool Description: Generates a random number within a specified range"
         in rendered
     )
     assert (
-        "Tool Name: Random Number Generator\nTool Arguments: {'min_value': {'description': 'The minimum value of the range (inclusive)', 'type': 'int'}, 'max_value': {'description': 'The maximum value of the range (inclusive)', 'type': 'int'}}\nTool Description: Generates a random number within a specified range"
+        'Tool Name: Random Number Generator\nTool Arguments: {"min_value": {"description": "The minimum value of the range (inclusive)", "type": "int"}, "max_value": {"description": "The maximum value of the range (inclusive)", "type": "int"}}\nTool Description: Generates a random number within a specified range'
         in rendered
     )