** Please make sure you read the contribution guide and file the issues in the right place. **
Contribution guide.

Describe the bug
The FunctionTool's parameter filtering logic incorrectly filters out parameters for functions that use **kwargs, causing CrewAI-style tools to receive empty parameter dictionaries. This breaks integration with CrewAI tools that use the *args, **kwargs pattern.

To Reproduce
Please share a minimal code and data to reproduce your problem.
Steps to reproduce the behavior:

"""Main agent for the speaker research system following ADK multi-agent patterns."""

from google.adk.agents import LlmAgent
from google.adk.tools import FunctionTool 
from google.adk.tools.crewai_tool import CrewaiTool
from crewai_tools import SerperDevTool

# Instantiate the CrewAI tool
serper_tool_instance = SerperDevTool()

sweb_search_tool = CrewaiTool(
    tool=serper_tool_instance,
    name='WebSearch',
    description='Searches the web for information.'
)

# Main root agent that coordinates the entire system
root_agent = LlmAgent(
    model="gemini-2.5-flash",
    name='research_agent',
    description='A specialized agent for researching and evaluating.',
    instruction='''You are a research agent. When you need to search for information use WebSearch tool.'',
    tools=[web_search_tool]
)

In src/google/adk/tools/function_tool.py line 109, the parameter filtering logic:
args_to_call = {k: v for k, v in args_to_call.items() if k in valid_params}
Only allows parameters that match the function's signature parameter names. For functions with **kwargs, the valid parameters are {"self", "args", "kwargs"}, so specific parameters like search_query get filtered out.

Expected behavior
A clear and concise description of what you expected to happen.

Functions that accept **kwargs should receive all provided parameters (except self and tool_context) in their kwargs dictionary.

Desktop (please complete the following information):

• OS: macOS
• Python version: 3.13
• ADK version: (1.15.1) [current main branch]

Model Information:

• Are you using LiteLLM: No
• Which model is being used: `gemini-2.5-flash