fix: propagate Codex reasoning effort

hindsight-api-slim/hindsight_api/engine/providers/codex_llm.py

@@ -172,8 +172,8 @@ def __init__(
         if self.model.startswith("openai/"):
             self.model = self.model[len("openai/") :]
 
-        # Map reasoning effort to Codex reasoning summary format
-        # Codex supports: "auto", "concise", "detailed"
+        # Reasoning summary controls presentation separately from the backend's
+        # reasoning effort, which is sent unchanged in each request payload.
         self.reasoning_summary = self._map_reasoning_effort(reasoning_effort)
 
         # HTTP client for SSE streaming
@@ -448,7 +448,7 @@ async def call(
             "tools": [],
             "tool_choice": "auto",
             "parallel_tool_calls": True,
-            "reasoning": {"summary": reasoning_summary},
+            "reasoning": {"effort": self.reasoning_effort, "summary": reasoning_summary},
             "store": False,  # Codex uses stateless mode
             "stream": True,  # SSE streaming
             "include": ["reasoning.encrypted_content"],
@@ -831,7 +831,7 @@ async def call_with_tools(
                 else tool_choice.mode.value
             ),
             "parallel_tool_calls": True,
-            "reasoning": {"summary": reasoning_summary},
+            "reasoning": {"effort": self.reasoning_effort, "summary": reasoning_summary},
             "store": False,
             "stream": True,
             "include": ["reasoning.encrypted_content"],

hindsight-api-slim/tests/test_codex_reasoning_effort.py

@@ -0,0 +1,64 @@
+"""Regression tests for Codex reasoning-effort request serialization."""
+
+from unittest.mock import AsyncMock, MagicMock, patch
+
+import pytest
+
+from hindsight_api.engine.providers.codex_llm import CodexLLM
+
+
+def build_llm(reasoning_effort: str = "high") -> CodexLLM:
+    with (
+        patch.object(CodexLLM, "_load_codex_auth", return_value=("token", "account")),
+        patch.object(CodexLLM, "_load_codex_refresh_token", return_value=None),
+    ):
+        return CodexLLM(
+            provider="openai-codex",
+            api_key="ignored",
+            base_url="https://chatgpt.com/backend-api",
+            model="gpt-5.6-luna",
+            reasoning_effort=reasoning_effort,
+        )
+
+
+@pytest.mark.asyncio
+async def test_call_sends_reasoning_effort_separately_from_summary() -> None:
+    llm = build_llm("high")
+    response = MagicMock()
+    response.raise_for_status.return_value = None
+
+    with (
+        patch.object(llm._client, "post", new_callable=AsyncMock) as mock_post,
+        patch.object(llm, "_parse_sse_stream", new_callable=AsyncMock, return_value="ok"),
+    ):
+        mock_post.return_value = response
+        await llm.call(messages=[{"role": "user", "content": "hello"}], max_retries=0)
+
+    assert mock_post.call_args.kwargs["json"]["reasoning"] == {
+        "effort": "high",
+        "summary": "detailed",
+    }
+
+
+@pytest.mark.asyncio
+async def test_call_with_tools_sends_reasoning_effort_separately_from_summary() -> None:
+    llm = build_llm("low")
+    response = MagicMock()
+    response.status_code = 200
+    response.raise_for_status.return_value = None
+
+    with (
+        patch.object(llm._client, "post", new_callable=AsyncMock) as mock_post,
+        patch.object(llm, "_parse_sse_tool_stream", new_callable=AsyncMock, return_value=(None, [])),
+    ):
+        mock_post.return_value = response
+        await llm.call_with_tools(
+            messages=[{"role": "user", "content": "hello"}],
+            tools=[],
+            max_retries=0,
+        )
+
+    assert mock_post.call_args.kwargs["json"]["reasoning"] == {
+        "effort": "low",
+        "summary": "concise",
+    }