Framework for orchestrating role-playing, autonomous AI agents. By fostering collaborative intelligence, CrewAI empowers agents to work together seamlessly, tackling complex tasks.
Enhance Gemini LLM Tool Handling and Add Test for Float Responses (#4273)
Lorenze Jay0cb40374de1f
diff --git a/lib/crewai/src/crewai/llms/providers/gemini/completion.py b/lib/crewai/src/crewai/llms/providers/gemini/completion.pyindex f3913c216..f208b49ee 100644--- a/lib/crewai/src/crewai/llms/providers/gemini/completion.py+++ b/lib/crewai/src/crewai/llms/providers/gemini/completion.py@@ -555,7 +555,11 @@ class GeminiCompletion(BaseLLM):response_data: dict[str, Any]try:- response_data = json.loads(text_content) if text_content else {}+ parsed = json.loads(text_content) if text_content else {}+ if isinstance(parsed, dict):+ response_data = parsed+ else:+ response_data = {"result": parsed}except (json.JSONDecodeError, TypeError):response_data = {"result": text_content}diff --git a/lib/crewai/tests/cassettes/llms/google/test_gemini_tool_returning_float.yaml b/lib/crewai/tests/cassettes/llms/google/test_gemini_tool_returning_float.yamlnew file mode 100644index 000000000..1047f77c3--- /dev/null+++ b/lib/crewai/tests/cassettes/llms/google/test_gemini_tool_returning_float.yaml@@ -0,0 +1,319 @@+interactions:+- request:+ body: '{"contents": [{"parts": [{"text": "\nCurrent Task: What is 10000 + 20000?+ Use the sum_numbers tool to calculate this.\n\nThis is the expected criteria+ for your final answer: The sum of the two numbers\nyou MUST return the actual+ complete content as the final answer, not a summary.\n\nThis is VERY important+ to you, your job depends on it!"}], "role": "user"}], "systemInstruction": {"parts":+ [{"text": "You are Calculator. You are a calculator that adds numbers.\nYour+ personal goal is: Calculate numbers accurately"}], "role": "user"}, "tools":+ [{"functionDeclarations": [{"description": "Add two numbers together and return+ the result", "name": "sum_numbers", "parameters": {"properties": {"a": {"description":+ "The first number to add", "title": "A", "type": "NUMBER"}, "b": {"description":+ "The second number to add", "title": "B", "type": "NUMBER"}}, "required": ["a",+ "b"], "type": "OBJECT"}}]}], "generationConfig": {"stopSequences": ["\nObservation:"]}}'+ headers:+ User-Agent:+ - X-USER-AGENT-XXX+ accept:+ - '*/*'+ accept-encoding:+ - ACCEPT-ENCODING-XXX+ connection:+ - keep-alive+ content-length:+ - '962'+ content-type:+ - application/json+ host:+ - generativelanguage.googleapis.com+ x-goog-api-client:+ - google-genai-sdk/1.49.0 gl-python/3.13.3+ x-goog-api-key:+ - X-GOOG-API-KEY-XXX+ method: POST+ uri: https://generativelanguage.googleapis.com/v1beta/models/gemini-2.0-flash-001:generateContent+ response:+ body:+ string: "{\n \"candidates\": [\n {\n \"content\": {\n \"parts\":+ [\n {\n \"functionCall\": {\n \"name\": \"sum_numbers\",\n+ \ \"args\": {\n \"a\": 10000,\n \"b\":+ 20000\n }\n }\n }\n ],\n \"role\":+ \"model\"\n },\n \"finishReason\": \"STOP\",\n \"avgLogprobs\":+ -0.00059548033667462211\n }\n ],\n \"usageMetadata\": {\n \"promptTokenCount\":+ 127,\n \"candidatesTokenCount\": 7,\n \"totalTokenCount\": 134,\n \"promptTokensDetails\":+ [\n {\n \"modality\": \"TEXT\",\n \"tokenCount\": 127\n+ \ }\n ],\n \"candidatesTokensDetails\": [\n {\n \"modality\":+ \"TEXT\",\n \"tokenCount\": 7\n }\n ]\n },\n \"modelVersion\":+ \"gemini-2.0-flash-001\",\n \"responseId\": \"bLBzabiACaP3-8YP7s-P6QI\"\n}\n"+ headers:+ Alt-Svc:+ - h3=":443"; ma=2592000,h3-29=":443"; ma=2592000+ Content-Type:+ - application/json; charset=UTF-8+ Date:+ - Fri, 23 Jan 2026 17:31:24 GMT+ Server:+ - scaffolding on HTTPServer2+ Server-Timing:+ - gfet4t7; dur=673+ Transfer-Encoding:+ - chunked+ Vary:+ - Origin+ - X-Origin+ - Referer+ X-Content-Type-Options:+ - X-CONTENT-TYPE-XXX+ X-Frame-Options:+ - X-FRAME-OPTIONS-XXX+ X-XSS-Protection:+ - '0'+ status:+ code: 200+ message: OK+- request:+ body: '{"contents": [{"parts": [{"text": "\nCurrent Task: What is 10000 + 20000?+ Use the sum_numbers tool to calculate this.\n\nThis is the expected criteria+ for your final answer: The sum of the two numbers\nyou MUST return the actual+ complete content as the final answer, not a summary.\n\nThis is VERY important+ to you, your job depends on it!"}], "role": "user"}, {"parts": [{"functionCall":+ {"args": {"a": 10000, "b": 20000}, "name": "sum_numbers"}}], "role": "model"},+ {"parts": [{"functionResponse": {"name": "sum_numbers", "response": {"result":+ 30000}}}], "role": "user"}, {"parts": [{"text": "Analyze the tool result. If+ requirements are met, provide the Final Answer. Otherwise, call the next tool.+ Deliver only the answer without meta-commentary."}], "role": "user"}], "systemInstruction":+ {"parts": [{"text": "You are Calculator. You are a calculator that adds numbers.\nYour+ personal goal is: Calculate numbers accurately"}], "role": "user"}, "tools":+ [{"functionDeclarations": [{"description": "Add two numbers together and return+ the result", "name": "sum_numbers", "parameters": {"properties": {"a": {"description":+ "The first number to add", "title": "A", "type": "NUMBER"}, "b": {"description":+ "The second number to add", "title": "B", "type": "NUMBER"}}, "required": ["a",+ "b"], "type": "OBJECT"}}]}], "generationConfig": {"stopSequences": ["\nObservation:"]}}'+ headers:+ User-Agent:+ - X-USER-AGENT-XXX+ accept:+ - '*/*'+ accept-encoding:+ - ACCEPT-ENCODING-XXX+ connection:+ - keep-alive+ content-length:+ - '1374'+ content-type:+ - application/json+ host:+ - generativelanguage.googleapis.com+ x-goog-api-client:+ - google-genai-sdk/1.49.0 gl-python/3.13.3+ x-goog-api-key:+ - X-GOOG-API-KEY-XXX+ method: POST+ uri: https://generativelanguage.googleapis.com/v1beta/models/gemini-2.0-flash-001:generateContent+ response:+ body:+ string: "{\n \"candidates\": [\n {\n \"content\": {\n \"parts\":+ [\n {\n \"text\": \"\"\n }\n ],\n \"role\":+ \"model\"\n },\n \"finishReason\": \"STOP\"\n }\n ],\n \"usageMetadata\":+ {\n \"promptTokenCount\": 171,\n \"totalTokenCount\": 171,\n \"promptTokensDetails\":+ [\n {\n \"modality\": \"TEXT\",\n \"tokenCount\": 171\n+ \ }\n ]\n },\n \"modelVersion\": \"gemini-2.0-flash-001\",\n \"responseId\":+ \"bLBzaaKgMc-ajrEPk7bIuQ8\"\n}\n"+ headers:+ Alt-Svc:+ - h3=":443"; ma=2592000,h3-29=":443"; ma=2592000+ Content-Type:+ - application/json; charset=UTF-8+ Date:+ - Fri, 23 Jan 2026 17:31:25 GMT+ Server:+ - scaffolding on HTTPServer2+ Server-Timing:+ - gfet4t7; dur=382+ Transfer-Encoding:+ - chunked+ Vary:+ - Origin+ - X-Origin+ - Referer+ X-Content-Type-Options:+ - X-CONTENT-TYPE-XXX+ X-Frame-Options:+ - X-FRAME-OPTIONS-XXX+ X-XSS-Protection:+ - '0'+ status:+ code: 200+ message: OK+- request:+ body: '{"contents": [{"parts": [{"text": "\nCurrent Task: What is 10000 + 20000?+ Use the sum_numbers tool to calculate this.\n\nThis is the expected criteria+ for your final answer: The sum of the two numbers\nyou MUST return the actual+ complete content as the final answer, not a summary.\n\nThis is VERY important+ to you, your job depends on it!"}], "role": "user"}, {"parts": [{"functionCall":+ {"args": {"a": 10000, "b": 20000}, "name": "sum_numbers"}}], "role": "model"},+ {"parts": [{"functionResponse": {"name": "sum_numbers", "response": {"result":+ 30000}}}], "role": "user"}, {"parts": [{"text": "Analyze the tool result. If+ requirements are met, provide the Final Answer. Otherwise, call the next tool.+ Deliver only the answer without meta-commentary."}], "role": "user"}, {"parts":+ [{"text": "\nCurrent Task: What is 10000 + 20000? Use the sum_numbers tool to+ calculate this.\n\nThis is the expected criteria for your final answer: The+ sum of the two numbers\nyou MUST return the actual complete content as the final+ answer, not a summary.\n\nThis is VERY important to you, your job depends on+ it!"}], "role": "user"}], "systemInstruction": {"parts": [{"text": "You are+ Calculator. You are a calculator that adds numbers.\nYour personal goal is:+ Calculate numbers accurately\n\nYou are Calculator. You are a calculator that+ adds numbers.\nYour personal goal is: Calculate numbers accurately"}], "role":+ "user"}, "tools": [{"functionDeclarations": [{"description": "Add two numbers+ together and return the result", "name": "sum_numbers", "parameters": {"properties":+ {"a": {"description": "The first number to add", "title": "A", "type": "NUMBER"},+ "b": {"description": "The second number to add", "title": "B", "type": "NUMBER"}},+ "required": ["a", "b"], "type": "OBJECT"}}]}], "generationConfig": {"stopSequences":+ ["\nObservation:"]}}'+ headers:+ User-Agent:+ - X-USER-AGENT-XXX+ accept:+ - '*/*'+ accept-encoding:+ - ACCEPT-ENCODING-XXX+ connection:+ - keep-alive+ content-length:+ - '1837'+ content-type:+ - application/json+ host:+ - generativelanguage.googleapis.com+ x-goog-api-client:+ - google-genai-sdk/1.49.0 gl-python/3.13.3+ x-goog-api-key:+ - X-GOOG-API-KEY-XXX+ method: POST+ uri: https://generativelanguage.googleapis.com/v1beta/models/gemini-2.0-flash-001:generateContent+ response:+ body:+ string: "{\n \"candidates\": [\n {\n \"content\": {\n \"parts\":+ [\n {\n \"text\": \"\"\n }\n ],\n \"role\":+ \"model\"\n },\n \"finishReason\": \"STOP\"\n }\n ],\n \"usageMetadata\":+ {\n \"promptTokenCount\": 271,\n \"totalTokenCount\": 271,\n \"promptTokensDetails\":+ [\n {\n \"modality\": \"TEXT\",\n \"tokenCount\": 271\n+ \ }\n ]\n },\n \"modelVersion\": \"gemini-2.0-flash-001\",\n \"responseId\":+ \"bbBzaczHDcW7jrEPgaj1CA\"\n}\n"+ headers:+ Alt-Svc:+ - h3=":443"; ma=2592000,h3-29=":443"; ma=2592000+ Content-Type:+ - application/json; charset=UTF-8+ Date:+ - Fri, 23 Jan 2026 17:31:25 GMT+ Server:+ - scaffolding on HTTPServer2+ Server-Timing:+ - gfet4t7; dur=410+ Transfer-Encoding:+ - chunked+ Vary:+ - Origin+ - X-Origin+ - Referer+ X-Content-Type-Options:+ - X-CONTENT-TYPE-XXX+ X-Frame-Options:+ - X-FRAME-OPTIONS-XXX+ X-XSS-Protection:+ - '0'+ status:+ code: 200+ message: OK+- request:+ body: '{"contents": [{"parts": [{"text": "\nCurrent Task: What is 10000 + 20000?+ Use the sum_numbers tool to calculate this.\n\nThis is the expected criteria+ for your final answer: The sum of the two numbers\nyou MUST return the actual+ complete content as the final answer, not a summary.\n\nThis is VERY important+ to you, your job depends on it!"}], "role": "user"}, {"parts": [{"functionCall":+ {"args": {"a": 10000, "b": 20000}, "name": "sum_numbers"}}], "role": "model"},+ {"parts": [{"functionResponse": {"name": "sum_numbers", "response": {"result":+ 30000}}}], "role": "user"}, {"parts": [{"text": "Analyze the tool result. If+ requirements are met, provide the Final Answer. Otherwise, call the next tool.+ Deliver only the answer without meta-commentary."}], "role": "user"}, {"parts":+ [{"text": "\nCurrent Task: What is 10000 + 20000? Use the sum_numbers tool to+ calculate this.\n\nThis is the expected criteria for your final answer: The+ sum of the two numbers\nyou MUST return the actual complete content as the final+ answer, not a summary.\n\nThis is VERY important to you, your job depends on+ it!"}], "role": "user"}, {"parts": [{"text": "\nCurrent Task: What is 10000+ + 20000? Use the sum_numbers tool to calculate this.\n\nThis is the expected+ criteria for your final answer: The sum of the two numbers\nyou MUST return+ the actual complete content as the final answer, not a summary.\n\nThis is VERY+ important to you, your job depends on it!"}], "role": "user"}], "systemInstruction":+ {"parts": [{"text": "You are Calculator. You are a calculator that adds numbers.\nYour+ personal goal is: Calculate numbers accurately\n\nYou are Calculator. You are+ a calculator that adds numbers.\nYour personal goal is: Calculate numbers accurately\n\nYou+ are Calculator. You are a calculator that adds numbers.\nYour personal goal+ is: Calculate numbers accurately"}], "role": "user"}, "tools": [{"functionDeclarations":+ [{"description": "Add two numbers together and return the result", "name": "sum_numbers",+ "parameters": {"properties": {"a": {"description": "The first number to add",+ "title": "A", "type": "NUMBER"}, "b": {"description": "The second number to+ add", "title": "B", "type": "NUMBER"}}, "required": ["a", "b"], "type": "OBJECT"}}]}],+ "generationConfig": {"stopSequences": ["\nObservation:"]}}'+ headers:+ User-Agent:+ - X-USER-AGENT-XXX+ accept:+ - '*/*'+ accept-encoding:+ - ACCEPT-ENCODING-XXX+ connection:+ - keep-alive+ content-length:+ - '2300'+ content-type:+ - application/json+ host:+ - generativelanguage.googleapis.com+ x-goog-api-client:+ - google-genai-sdk/1.49.0 gl-python/3.13.3+ x-goog-api-key:+ - X-GOOG-API-KEY-XXX+ method: POST+ uri: https://generativelanguage.googleapis.com/v1beta/models/gemini-2.0-flash-001:generateContent+ response:+ body:+ string: "{\n \"candidates\": [\n {\n \"content\": {\n \"parts\":+ [\n {\n \"text\": \"\\n{\\\"sum_numbers_response\\\":+ {\\\"result\\\": 30000}}\\n\"\n }\n ],\n \"role\":+ \"model\"\n },\n \"finishReason\": \"STOP\",\n \"avgLogprobs\":+ -0.0038021293125654523\n }\n ],\n \"usageMetadata\": {\n \"promptTokenCount\":+ 371,\n \"candidatesTokenCount\": 19,\n \"totalTokenCount\": 390,\n \"promptTokensDetails\":+ [\n {\n \"modality\": \"TEXT\",\n \"tokenCount\": 371\n+ \ }\n ],\n \"candidatesTokensDetails\": [\n {\n \"modality\":+ \"TEXT\",\n \"tokenCount\": 19\n }\n ]\n },\n \"modelVersion\":+ \"gemini-2.0-flash-001\",\n \"responseId\": \"bbBzaauxJ_SgjrEP7onK2Ak\"\n}\n"+ headers:+ Alt-Svc:+ - h3=":443"; ma=2592000,h3-29=":443"; ma=2592000+ Content-Type:+ - application/json; charset=UTF-8+ Date:+ - Fri, 23 Jan 2026 17:31:26 GMT+ Server:+ - scaffolding on HTTPServer2+ Server-Timing:+ - gfet4t7; dur=454+ Transfer-Encoding:+ - chunked+ Vary:+ - Origin+ - X-Origin+ - Referer+ X-Content-Type-Options:+ - X-CONTENT-TYPE-XXX+ X-Frame-Options:+ - X-FRAME-OPTIONS-XXX+ X-XSS-Protection:+ - '0'+ status:+ code: 200+ message: OK+version: 1diff --git a/lib/crewai/tests/llms/google/test_google.py b/lib/crewai/tests/llms/google/test_google.pyindex 619f1e7c6..33d53da2d 100644--- a/lib/crewai/tests/llms/google/test_google.py+++ b/lib/crewai/tests/llms/google/test_google.py@@ -635,6 +635,54 @@ def test_gemini_token_usage_tracking():assert usage.total_tokens > 0+def test_gemini_tool_returning_float():+ """+ Test that Gemini properly handles tools that return non-dict values like floats.++ This is an end-to-end test that verifies the agent can use a tool that returns+ a float (which gets wrapped in {"result": value} for Gemini's FunctionResponse).+ """+ from pydantic import BaseModel, Field+ from typing import Type+ from crewai.tools import BaseTool++ class SumNumbersToolInput(BaseModel):+ a: float = Field(..., description="The first number to add")+ b: float = Field(..., description="The second number to add")++ class SumNumbersTool(BaseTool):+ name: str = "sum_numbers"+ description: str = "Add two numbers together and return the result"+ args_schema: Type[BaseModel] = SumNumbersToolInput++ def _run(self, a: float, b: float) -> float:+ return a + b++ sum_tool = SumNumbersTool()++ agent = Agent(+ role="Calculator",+ goal="Calculate numbers accurately",+ backstory="You are a calculator that adds numbers.",+ llm=LLM(model="google/gemini-2.0-flash-001"),+ tools=[sum_tool],+ verbose=True,+ )++ task = Task(+ description="What is 10000 + 20000? Use the sum_numbers tool to calculate this.",+ expected_output="The sum of the two numbers",+ agent=agent,+ )++ crew = Crew(agents=[agent], tasks=[task], verbose=True)+ result = crew.kickoff()++ # The result should contain 30000 (the sum)+ assert "30000" in result.raw++def test_gemini_stop_sequences_sync():"""Test that stop and stop_sequences attributes stay synchronized."""llm = LLM(model="google/gemini-2.0-flash-001")