fix(kimi): don't send Anthropic thinking to api.kimi.com/coding (#13826)

Kimi's /coding endpoint speaks the Anthropic Messages protocol but has
its own thinking semantics: when thinking.enabled is sent, Kimi validates
the history and requires every prior assistant tool-call message to carry
OpenAI-style reasoning_content. The Anthropic path never populates that
field, and convert_messages_to_anthropic strips Anthropic thinking blocks
on third-party endpoints — so after one tool-calling turn the next request
fails with:

  HTTP 400: thinking is enabled but reasoning_content is missing in
  assistant tool call message at index N

Kimi on chat_completions handles thinking via extra_body in
ChatCompletionsTransport (#13503). On the Anthropic route, drop the
parameter entirely and let Kimi drive reasoning server-side.

build_anthropic_kwargs now gates the reasoning_config -> thinking block
on not _is_kimi_coding_endpoint(base_url).

Tests: 8 new parametric tests cover /coding, /coding/v1, /coding/anthropic,
/coding/ (trailing slash), explicit disabled, other third-party endpoints
still getting thinking (MiniMax), native Anthropic unaffected, and the
non-/coding Kimi root route.

agent/anthropic_adapter.py

@@ -1426,11 +1426,25 @@ def build_anthropic_kwargs(
     # MiniMax Anthropic-compat endpoints support thinking (manual mode only,
     # not adaptive).  Haiku does NOT support extended thinking — skip entirely.
     #
+    # Kimi's /coding endpoint speaks the Anthropic Messages protocol but has
+    # its own thinking semantics: when ``thinking.enabled`` is sent, Kimi
+    # validates the message history and requires every prior assistant
+    # tool-call message to carry OpenAI-style ``reasoning_content``.  The
+    # Anthropic path never populates that field, and
+    # ``convert_messages_to_anthropic`` strips all Anthropic thinking blocks
+    # on third-party endpoints — so the request fails with HTTP 400
+    # "thinking is enabled but reasoning_content is missing in assistant
+    # tool call message at index N".  Kimi's reasoning is driven server-side
+    # on the /coding route, so skip Anthropic's thinking parameter entirely
+    # for that host.  (Kimi on chat_completions enables thinking via
+    # extra_body in the ChatCompletionsTransport — see #13503.)
+    #
     # On 4.7+ the `thinking.display` field defaults to "omitted", which
     # silently hides reasoning text that Hermes surfaces in its CLI. We
     # request "summarized" so the reasoning blocks stay populated — matching
     # 4.6 behavior and preserving the activity-feed UX during long tool runs.
-    if reasoning_config and isinstance(reasoning_config, dict):
+    _is_kimi_coding = _is_kimi_coding_endpoint(base_url)
+    if reasoning_config and isinstance(reasoning_config, dict) and not _is_kimi_coding:
         if reasoning_config.get("enabled") is not False and "haiku" not in model.lower():
             effort = str(reasoning_config.get("effort", "medium")).lower()
             budget = THINKING_BUDGET.get(effort, 8000)

tests/agent/test_kimi_coding_anthropic_thinking.py

@@ -0,0 +1,115 @@
+"""Regression guard: don't send Anthropic ``thinking`` to Kimi's /coding endpoint.
+
+Kimi's ``api.kimi.com/coding`` endpoint speaks the Anthropic Messages protocol
+but has its own thinking semantics.  When ``thinking.enabled`` is present in
+the request, Kimi validates the message history and requires every prior
+assistant tool-call message to carry OpenAI-style ``reasoning_content``.
+
+The Anthropic path never populates that field, and
+``convert_messages_to_anthropic`` strips Anthropic thinking blocks on
+third-party endpoints — so after one turn with tool calls the next request
+fails with HTTP 400::
+
+    thinking is enabled but reasoning_content is missing in assistant
+    tool call message at index N
+
+Kimi on the chat_completions route handles ``thinking`` via ``extra_body`` in
+``ChatCompletionsTransport`` (#13503).  On the Anthropic route the right
+thing to do is drop the parameter entirely and let Kimi drive reasoning
+server-side.
+"""
+
+from __future__ import annotations
+
+import pytest
+
+
+class TestKimiCodingSkipsAnthropicThinking:
+    """build_anthropic_kwargs must not inject ``thinking`` for Kimi /coding."""
+
+    @pytest.mark.parametrize(
+        "base_url",
+        [
+            "https://api.kimi.com/coding",
+            "https://api.kimi.com/coding/v1",
+            "https://api.kimi.com/coding/anthropic",
+            "https://api.kimi.com/coding/",
+        ],
+    )
+    def test_kimi_coding_endpoint_omits_thinking(self, base_url: str) -> None:
+        from agent.anthropic_adapter import build_anthropic_kwargs
+
+        kwargs = build_anthropic_kwargs(
+            model="kimi-k2.5",
+            messages=[{"role": "user", "content": "hello"}],
+            tools=None,
+            max_tokens=4096,
+            reasoning_config={"enabled": True, "effort": "medium"},
+            base_url=base_url,
+        )
+        assert "thinking" not in kwargs, (
+            "Anthropic thinking must not be sent to Kimi /coding — "
+            "endpoint requires reasoning_content on history we don't preserve."
+        )
+        assert "output_config" not in kwargs
+
+    def test_kimi_coding_with_explicit_disabled_also_omits(self) -> None:
+        from agent.anthropic_adapter import build_anthropic_kwargs
+
+        kwargs = build_anthropic_kwargs(
+            model="kimi-k2.5",
+            messages=[{"role": "user", "content": "hello"}],
+            tools=None,
+            max_tokens=4096,
+            reasoning_config={"enabled": False},
+            base_url="https://api.kimi.com/coding",
+        )
+        assert "thinking" not in kwargs
+
+    def test_non_kimi_third_party_still_gets_thinking(self) -> None:
+        """MiniMax and other third-party Anthropic endpoints must retain thinking."""
+        from agent.anthropic_adapter import build_anthropic_kwargs
+
+        kwargs = build_anthropic_kwargs(
+            model="MiniMax-M2.7",
+            messages=[{"role": "user", "content": "hello"}],
+            tools=None,
+            max_tokens=4096,
+            reasoning_config={"enabled": True, "effort": "medium"},
+            base_url="https://api.minimax.io/anthropic",
+        )
+        assert "thinking" in kwargs
+        assert kwargs["thinking"]["type"] == "enabled"
+
+    def test_native_anthropic_still_gets_thinking(self) -> None:
+        from agent.anthropic_adapter import build_anthropic_kwargs
+
+        kwargs = build_anthropic_kwargs(
+            model="claude-sonnet-4-20250514",
+            messages=[{"role": "user", "content": "hello"}],
+            tools=None,
+            max_tokens=4096,
+            reasoning_config={"enabled": True, "effort": "medium"},
+            base_url=None,
+        )
+        assert "thinking" in kwargs
+
+    def test_kimi_root_endpoint_unaffected(self) -> None:
+        """Only the /coding route is special-cased — plain api.kimi.com is not.
+
+        ``api.kimi.com`` without ``/coding`` uses the chat_completions transport
+        (see runtime_provider._detect_api_mode_for_url); build_anthropic_kwargs
+        should never see it, but if it somehow does we should not suppress
+        thinking there — that path has different semantics.
+        """
+        from agent.anthropic_adapter import build_anthropic_kwargs
+
+        kwargs = build_anthropic_kwargs(
+            model="kimi-k2.5",
+            messages=[{"role": "user", "content": "hello"}],
+            tools=None,
+            max_tokens=4096,
+            reasoning_config={"enabled": True, "effort": "medium"},
+            base_url="https://api.kimi.com/v1",
+        )
+        assert "thinking" in kwargs