From 3122125fd0ce183d066bb8d9109f3e5496e32ecf Mon Sep 17 00:00:00 2001
From: Rudy Celekli <rudy@gradiahq.com>
Date: Mon, 12 Oct 2026 21:16:53 -0400
Subject: [PATCH] fix(agent): normalize null thinking text before Anthropic
 replay

---
 apps/agent/src/anthropic_compat.py            |  6 +--
 ...test_anthropic_null_thinking_regression.py | 41 +++++++++++++++++++
 2 files changed, 44 insertions(+), 3 deletions(-)
 create mode 100644 apps/agent/tests/test_anthropic_null_thinking_regression.py

diff --git a/apps/agent/src/anthropic_compat.py b/apps/agent/src/anthropic_compat.py
index 5417650..17f2b1c 100644
--- a/apps/agent/src/anthropic_compat.py
+++ b/apps/agent/src/anthropic_compat.py
@@ -7,7 +7,7 @@
    for OpenAI, rejected by langchain-anthropic.
 2. Streaming a zero-length thinking block accumulates {type, signature} with
    no "thinking" key; Anthropic 400s on replay in the tool loop unless the
-   empty text field is restored."""
+   empty text field is restored. A persisted null text value needs the same repair."""
 
 from langchain.agents.middleware import AgentMiddleware, ModelRequest
 from langchain_core.messages import AIMessage
@@ -19,13 +19,13 @@ def _repaired_message(message):
     if not isinstance(message, AIMessage) or not isinstance(message.content, list):
         return message
     if not any(
-        isinstance(b, dict) and b.get("type") == "thinking" and "thinking" not in b
+        isinstance(b, dict) and b.get("type") == "thinking" and b.get("thinking") is None
         for b in message.content
     ):
         return message
     content = [
         {**b, "thinking": ""}
-        if isinstance(b, dict) and b.get("type") == "thinking" and "thinking" not in b
+        if isinstance(b, dict) and b.get("type") == "thinking" and b.get("thinking") is None
         else b
         for b in message.content
     ]
diff --git a/apps/agent/tests/test_anthropic_null_thinking_regression.py b/apps/agent/tests/test_anthropic_null_thinking_regression.py
new file mode 100644
index 0000000..c876be8
--- /dev/null
+++ b/apps/agent/tests/test_anthropic_null_thinking_regression.py
@@ -0,0 +1,41 @@
+import asyncio
+
+import pytest
+from anthropic.types.thinking_block import ThinkingBlock
+from langchain.agents.middleware import ModelRequest
+from langchain_anthropic import ChatAnthropic
+from langchain_core.messages import AIMessage, HumanMessage
+
+from src.anthropic_compat import ConsecutiveSystemMessagesMiddleware
+
+
+@pytest.mark.parametrize("use_async", [False, True])
+def test_null_thinking_replays_as_valid_sdk_text_without_mutating_history(use_async):
+    model = ChatAnthropic(model="claude-sonnet-4-5-20250929", api_key="test-key")
+    original = AIMessage(id="replayed", content=[
+        {"type": "thinking", "thinking": None, "signature": "signature", "cache_control": {"type": "ephemeral"}},
+        {"type": "text", "text": "Ready"},
+    ])
+    request = ModelRequest(model=model, messages=[HumanMessage(content="Continue"), original])
+    middleware = ConsecutiveSystemMessagesMiddleware()
+
+    def verify(repaired):
+        payload = model._get_request_payload(repaired.messages)
+        block = payload["messages"][-1]["content"][0]
+        ThinkingBlock.model_validate(block)
+        assert block["thinking"] == ""
+        assert block["signature"] == "signature"
+        assert block["cache_control"] == {"type": "ephemeral"}
+        assert repaired.messages[-1].id == "replayed"
+        assert repaired.messages[-1].content[1] == original.content[1]
+        return "verified"
+
+    async def averify(repaired):
+        return verify(repaired)
+
+    if use_async:
+        result = asyncio.run(middleware.awrap_model_call(request, averify))
+    else:
+        result = middleware.wrap_model_call(request, verify)
+    assert result == "verified"
+    assert original.content[0]["thinking"] is None