LangchainTool.from_config blanks the name and description a config leaves out #7900

🔴 Required Information

Describe the Bug:
LangchainToolConfig defaults name and description to '' (src/google/adk/integrations/langchain/langchain_tool.py:211,214), and LangchainTool.__init__ treats anything other than None as an override (:110,117). So a YAML agent that sets only tool gets a tool with an empty name and an empty description, and a config that sets only one of them blanks the other.

Two such tools in one agent both declare name=''. The second replaces the first in tools_dict, so the first can no longer be called, and the model is told about two functions with no name and no description.

docs/guides/integrations/langchain/langchain_tool/index.md describes both options as overrides of the Langchain tool's own metadata. CrewaiTool, whose config has the same '' defaults, falls back to the wrapped tool's values (src/google/adk/integrations/crewai/crewai_tool.py:52,58). Calling LangchainTool(tool) directly is not affected, because both parameters default to None there.

Steps to Reproduce:

1. Install google-adk from main at 8cd3b12c with langchain-core (the v2.13.0 tag has the same code).
2. Run the script under "Minimal Reproduction Code". It makes no model call and needs no API key.

Expected Behavior:
Options the config leaves out keep the wrapped tool's name and description:

[('add', 'Adds two numbers.'), ('subtract', 'Subtracts b from a.')]

Observed Behavior:

[('', ''), ('', '')]

Environment Details:

• ADK Library Version (pip show google-adk): 2.13.0, main at 8cd3b12c
• Desktop OS: Windows 11
• Python Version (python -V): Python 3.11.3

Model Information:

• Are you using LiteLLM: No
• Which model is being used: N/A, the script makes no model call (the Runner run under "Logs" uses a scripted model)

🟡 Optional Information

Regression:
No. The '' defaults and the is not None checks have both been there since from_config was added in 1686cc5.

Logs:
The same kind of agent through InMemoryRunner, with a scripted model that records the request it receives:

WARNING: Duplicate tool name '': the previously registered tool is shadowed and can no longer be called.
tools: [('', ''), ('', '')]
declarations sent to the model:
  name='' description=''
  name='' description=''
tools_dict keys: ['']

Screenshots / Video:
N/A

Additional Context:
No test calls LangchainTool.from_config today.

Minimal Reproduction Code:

import pathlib
import tempfile

from google.adk.agents import config_agent_utils
from langchain_core.tools import tool


@tool
def add(a: int, b: int) -> int:
  """Adds two numbers."""
  return a + b


@tool
def subtract(a: int, b: int) -> int:
  """Subtracts b from a."""
  return a - b


CONFIG = """\
name: math_agent
model: gemini-2.5-flash
instruction: Do the maths.
tools:
  - name: google.adk.integrations.langchain.LangchainTool
    args:
      tool: __main__.add
  - name: google.adk.integrations.langchain.LangchainTool
    args:
      tool: __main__.subtract
"""

with tempfile.TemporaryDirectory() as folder:
  path = pathlib.Path(folder, "root_agent.yaml")
  path.write_text(CONFIG)
  agent = config_agent_utils.from_config(str(path))

print([(t.name, t.description) for t in agent.tools])

How often has this issue occurred?:

• Always (100%)