Sentry LLM Monitoring does not capture tool use or thinking for Anthropic

[DanKing1903]

When capturing LLM outputs sentry should capture tool use blocks and thinking blocks as part of the response object, this is vital information for developers working on LLM products. Currently only text responses are captured.

sentry-python/sentry_sdk/integrations/anthropic.py

Lines 78 to 85 in adcfa0f

	if hasattr(item, "text"):
	responses.append(
	{
	"type": item.type,
	"text": item.text,
	}
	)
	return responses

https://docs.anthropic.com/en/api/messages#response-content