Bug: Thinking blocks not preserved in Anthropic conversations with tool calls

[habema]

Error Details:

litellm.BadRequestError: AnthropicException - b'{"type":"error","error":{"type":"invalid_request_error","message":"messages.1.content.0.type: Expected `thinking` or `redacted_thinking`, but found `tool_use`. When `thinking` is enabled, a final `assistant` message must start with a thinking block (preceeding the lastmost set of `tool_use` and `tool_result` blocks). We recommend you include thinking blocks from previous turns. To avoid this requirement, disable `thinking`. Please consult our documentation at https://docs.anthropic.com/en/docs/build-with-claude/extended-thinking"}}'

Root cause:

The SDK drops reasoning items when converting conversation history to API messages, violating Anthropic's requirement that assistant messages with tool calls must start with thinking blocks.

Minimal repro:

import asyncio
from agents import Agent, ModelSettings, Runner, SQLiteSession, function_tool
from openai.types.shared.reasoning import Reasoning

@function_tool
def get_weather(location: str) -> str:
    return f"The weather in {location} is sunny and 72°F."

async def main():
    agent = Agent(
        name="test_agent",
        model="litellm/anthropic/claude-sonnet-4-20250514",
        model_settings=ModelSettings(reasoning=Reasoning(effort="medium")),
        tools=[get_weather],
        instructions="You are a helpful assistant. When you need to use tools, think through the problem first.",
    )
    
    session = SQLiteSession(session_id="test_thinking_blocks")
    
    # This fails with the error above
    result = await Runner.run(
        agent, "What's the weather like in San Francisco?", session=session
    )

if __name__ == "__main__":
    asyncio.run(main())

Relevant Docs:

LiteLLM, Anthropic