test(e2e): check the tool input on the messages path too

The /v1/messages validator checked a tool_use block's name and id but not its
input, so a block whose location came back empty or wrong still passed, while
the chat side rejected the same damage. That gap predates this branch; it is
worth closing here because the point of the change is that every parallel call
is checked rather than counted.

AnthropicContentBlock now declares input as a typed field. It already survived
on extra="allow", but reaching it from a test needs a real field to keep the
e2e basedpyright gate at zero. Serialization is unchanged: bodies are dumped
with exclude_none, so a block without an input still replays exactly as before.

tests/e2e/llm_translation/test_together_ai_e2e.py

@@ -483,6 +483,9 @@ def _tool_use_blocks(content: list[AnthropicContentBlock] | None) -> list[Anthro
 def _validated_tool_use_id(block: AnthropicContentBlock) -> str:
     assert block.name == "get_weather", f"wrong tool called: {block}"
     assert block.id, f"tool_use block carries no id, so a tool_result cannot answer it: {block}"
+    assert block.input is not None, f"tool_use block carries no input: {block}"
+    args = _WeatherArgs.model_validate(block.input)
+    assert "paris" in args.location.lower(), f"tool input lost the location: {args}"
     return block.id
 
 

tests/e2e/models.py

@@ -421,6 +421,7 @@ class AnthropicContentBlock(BaseModel):
     text: str | None = None
     id: str | None = None
     name: str | None = None
+    input: dict[str, object] | None = None
 
 
 class AnthropicToolResultBlock(BaseModel):