Annotated type hint metadata not preserved in function tool declarations/tool schema

[MarlzRana]

🔴Required Information

Describe the Bug:
Annotated is the standard way to provide metadata about input parameters and output in a tool, most importantly descriptions and constraints. However, these today are stripped by the current logic before calling the Gemini API, so the underlying model never sees them. This is fixed in PR: #6879

But when using Annotated, parameter and response descriptions are missing.

Steps to Reproduce:

1. Enable the JSON_SCHEMA_FOR_FUNC_DECL feature (where we expect a complete tool schema to be made out of the parameter list)
2. Define a tool like the below

from typing import Annotated
from uuid import UUID

from pydantic import Field

def get_forecast(
    city_id: Annotated[
        UUID, Field(description="The city UUID as from the Google Places API")
    ],
    days: Annotated[
        int, Field(description="Number of days to forecast", ge=1, le=14)
    ],
) -> str:
  """Use when you want a multi-day weather forecast for a US city."""
  return ...

3. Set a breakpoint in src/google/adk/flows/llm_flows/base_llm_flow.py, in _call_llm_async (inside _call_llm_with_tracing, just before llm.generate_content_async(llm_request=llm_request, ...)) and inspect llm_request.config.tools[0].function_declarations[0].parameters_json_schema

Expected Behavior:
Observe that the Gemini API call tool schema is:

{
  "type": "object",
  "title": "get_forecastParams",
  "properties": {
    "city_id": {
      "type": "string",
      "format": "uuid",
      "title": "City Id",
      "description": "The city UUID as from the Google Places API"
    },
    "days": {
      "type": "integer",
      "minimum": 1,
      "maximum": 14,
      "title": "Days",
      "description": "Number of days to forecast"
    }
  },
  "required": ["city_id", "days"]
}

Where we see the tool description included.

Observed Behavior:

{
  "type": "object",
  "title": "get_forecastParams",
  "properties": {
    "city_id": {
      "type": "string",
      "format": "uuid",
      "title": "City Id"
    },
    "days": {
      "type": "integer",
      "title": "Days"
    }
  },
  "required": ["city_id", "days"]
}

Where we can see the tool description omitted.

Environment Details:

• ADK Library Version: 2.7.1
• Desktop OS: MacOS
• Python Version: 3.13

Model Information:

• Are you using LiteLLM: No
• Which model is being used: gemini-3.7-flash

---

🟡 Optional Information

Regression:
Not a regression.

How often has this issue occurred?:

• Always (100%)