fix(provider): stabilize Codex prompt cache routing (#5540)

nanobot/agent/runner.py

@@ -495,6 +495,7 @@ class AgentRunner:
             model=spec.runtime.model,
             messages=messages,
             state=spec.provider_state,
+            session_id=spec.session_key,
         )
         governance_config = ContextGovernanceConfig(
             provider=spec.runtime.provider,

nanobot/providers/base.py

@@ -252,10 +252,13 @@ class ProviderCallContext:
     The regular ``chat`` contract stays provider-agnostic. Responses-capable
     providers consume this context through the opt-in ``chat_with_context``
     hooks, while every other provider inherits the context-free delegation.
+    ``session_id`` gives providers a stable conversation-scoped routing key
+    without exposing that identity in the public message transcript.
     """
 
     conversation_state: ProviderConversationState | None = field(default=None, repr=False)
     context_window_tokens: int | None = None
+    session_id: str | None = field(default=None, repr=False)
 
 
 @dataclass(frozen=True, slots=True)
@@ -1640,6 +1643,7 @@ class LLMProvider(ABC):
                             context_window_tokens=(
                                 provider_context.context_window_tokens
                             ),
+                            session_id=provider_context.session_id,
                         )
                 if stripped is not None or stripped_context is not None:
                     logger.warning(

nanobot/providers/conversation_state.py

@@ -42,9 +42,11 @@ class ProviderConversationStateController:
         model: str | None,
         messages: list[dict[str, Any]],
         state: ProviderConversationState | None = None,
+        session_id: str | None = None,
     ) -> None:
         self._provider = provider
         self._model = model
+        self._session_id = session_id
         self._state = (
             state
             if state is not None
@@ -60,9 +62,12 @@ class ProviderConversationStateController:
         context_window_tokens: int | None,
     ) -> ProviderCallContext | None:
         """Return typed provider context for a request that does not resume state."""
-        if context_window_tokens is None:
+        if context_window_tokens is None and self._session_id is None:
             return None
-        return ProviderCallContext(context_window_tokens=context_window_tokens)
+        return ProviderCallContext(
+            context_window_tokens=context_window_tokens,
+            session_id=self._session_id,
+        )
 
     def prepare_request(
         self,
@@ -112,6 +117,7 @@ class ProviderConversationStateController:
                 if independent_context is not None
                 else None
             ),
+            session_id=self._session_id,
         )
 
     def observe_response(

nanobot/providers/fallback_provider.py

@@ -186,6 +186,7 @@ class FallbackProvider(LLMProvider):
         return ProviderCallContext(
             conversation_state=provider_context.conversation_state,
             context_window_tokens=context_window_tokens,
+            session_id=provider_context.session_id,
         )
 
     def _primary_available(self) -> bool:
@@ -541,6 +542,7 @@ class FallbackProvider(LLMProvider):
                 fallback_kwargs["provider_context"] = ProviderCallContext(
                     conversation_state=state,
                     context_window_tokens=context_window_tokens,
+                    session_id=provider_context.session_id,
                 )
             if fallback.reasoning_effort is None:
                 fallback_kwargs.pop("reasoning_effort", None)

nanobot/providers/openai_codex_provider.py

@@ -103,6 +103,7 @@ class OpenAICodexProvider(LLMProvider):
             provider=self._responses_state_provider(),
             model=_strip_model_prefix(model),
         )
+        session_id = provider_context.session_id if provider_context is not None else None
 
         body: dict[str, Any] = {
             "model": _strip_model_prefix(model),
@@ -111,10 +112,11 @@ class OpenAICodexProvider(LLMProvider):
             "instructions": system_prompt,
             "input": input_items,
             "text": {"verbosity": "medium"},
-            "prompt_cache_key": _prompt_cache_key(messages[:2]),
             "tool_choice": tool_choice or "auto",
             "parallel_tool_calls": True,
         }
+        if session_id:
+            body["prompt_cache_key"] = _prompt_cache_key(session_id)
         body["include"] = ["reasoning.encrypted_content"]
         reasoning_options = _build_reasoning_options(reasoning_effort)
         if replayed and "gpt-5.6" in _strip_model_prefix(model).lower():
@@ -496,9 +498,8 @@ async def _request_codex(
             return result
 
 
-def _prompt_cache_key(messages: list[dict[str, Any]]) -> str:
+def _prompt_cache_key(session_id: str) -> str:
-    raw = json.dumps(messages, ensure_ascii=True, sort_keys=True)
+    return hashlib.sha256(session_id.encode("utf-8")).hexdigest()
-    return hashlib.sha256(raw.encode("utf-8")).hexdigest()
 
 
 def _friendly_error(status_code: int, raw: str) -> str:

tests/agent/test_runner_fallback.py

@@ -384,12 +384,16 @@ class TestFallbackOnPrimaryError:
             messages=[{"role": "user", "content": "hi"}],
             model="gpt-5.6",
             max_tokens=10_000,
-            provider_context=ProviderCallContext(context_window_tokens=50_000),
+            provider_context=ProviderCallContext(
+                context_window_tokens=50_000,
+                session_id="webui:cache-test",
+            ),
         )
 
         primary_context = primary.context_calls[0]
         assert primary_context is not None
         assert primary_context.context_window_tokens == 200_000
+        assert primary_context.session_id == "webui:cache-test"
         assert resolve_compact_threshold(
             primary_context.context_window_tokens,
             10_000,

tests/agent/test_runner_runtime_identity.py

@@ -8,6 +8,7 @@ from nanobot.providers.base import (
     GenerationSettings,
     LLMProvider,
     LLMResponse,
+    ProviderCallContext,
     ToolCallRequest,
 )
 from nanobot.utils.llm_runtime import LLMRuntime
@@ -22,6 +23,7 @@ async def test_active_run_keeps_provider_captured_at_admission() -> None:
     first_calls = 0
     second_calls = 0
     request_temperatures: list[float] = []
+    request_session_ids: list[str | None] = []
     selected_runtime = LLMRuntime.capture(
         first_provider,
         "captured-model",
@@ -33,6 +35,9 @@ async def test_active_run_keeps_provider_captured_at_admission() -> None:
         nonlocal first_calls, selected_runtime
         first_calls += 1
         request_temperatures.append(kwargs["temperature"])
+        provider_context = kwargs["provider_context"]
+        assert isinstance(provider_context, ProviderCallContext)
+        request_session_ids.append(provider_context.session_id)
         selected_runtime = LLMRuntime.capture(
             second_provider,
             "future-model",
@@ -63,9 +68,11 @@ async def test_active_run_keeps_provider_captured_at_admission() -> None:
         runtime=selected_runtime,
         max_iterations=2,
         max_tool_result_chars=AgentDefaults().max_tool_result_chars,
+        session_key="webui:cache-test",
     ))
 
     assert first_calls == 2
     assert second_calls == 0
     assert request_temperatures == [0.2, 0.2]
+    assert request_session_ids == ["webui:cache-test", "webui:cache-test"]
     assert selected_runtime.provider is second_provider

tests/providers/test_conversation_state.py

@@ -9,6 +9,7 @@ import pytest
 from nanobot.providers.base import (
     LLMProvider,
     LLMResponse,
+    ProviderCallContext,
     ProviderConversationState,
     ToolCallRequest,
 )
@@ -289,3 +290,23 @@ def test_independent_request_exposes_context_without_capability_check() -> None:
     assert provider_context.conversation_state is None
     assert provider_context.context_window_tokens == 200_000
     provider.supports_native_compaction.assert_not_called()
+
+
+def test_independent_request_exposes_session_id_without_token_budget() -> None:
+    provider = _provider(compact=False)
+    messages = [{"role": "user", "content": "hello"}]
+    controller = ProviderConversationStateController(
+        provider=provider,
+        model="gpt-5.6",
+        messages=messages,
+        session_id="webui:cache-test",
+    )
+
+    provider_context = controller.prepare_request(
+        messages,
+        context_window_tokens=None,
+    )
+
+    assert provider_context == ProviderCallContext(
+        session_id="webui:cache-test",
+    )

tests/providers/test_openai_codex_provider.py

@@ -260,8 +260,8 @@ async def test_codex_request_uses_configured_proxy(monkeypatch) -> None:
 
 
 @pytest.mark.asyncio
-async def test_codex_prompt_cache_key_uses_stable_conversation_prefix(monkeypatch) -> None:
+async def test_codex_omits_prompt_cache_key_without_session_id(monkeypatch) -> None:
-    bodies: list[dict] = []
+    bodies: list[dict[str, Any]] = []
 
     _mock_codex_token(monkeypatch)
 
@@ -289,25 +289,40 @@ async def test_codex_prompt_cache_key_uses_stable_conversation_prefix(monkeypatc
             {"role": "assistant", "content": "first answer"},
         ],
     )
+
+    assert "prompt_cache_key" not in bodies[0]
+    assert "service_tier" not in bodies[0]
+
+
+@pytest.mark.asyncio
+async def test_codex_prompt_cache_key_prefers_stable_session_id(monkeypatch) -> None:
+    bodies: list[dict[str, Any]] = []
+    _mock_codex_token(monkeypatch)
+
+    async def fake_request(_url, _headers, body, **_kwargs):
+        bodies.append(body)
+        return provider_base.LLMResponse(content="ok")
+
+    monkeypatch.setattr("nanobot.providers.openai_codex_provider._request_codex", fake_request)
+    provider = OpenAICodexProvider()
+
+    for session_id, first_request in (
+        ("session-a", "first request"),
+        ("session-a", "different visible prefix"),
+        ("session-b", "first request"),
+    ):
         await provider.chat(
             [
                 {"role": "system", "content": "You are nanobot."},
-            {"role": "user", "content": "first request"},
+                {"role": "user", "content": first_request},
-            {"role": "assistant", "content": "first answer"},
-            {"role": "user", "content": "follow up"},
-        ],
-    )
-    await provider.chat(
-        [
-            {"role": "system", "content": "You are nanobot."},
-            {"role": "user", "content": "different request"},
-            {"role": "assistant", "content": "first answer"},
             ],
+            provider_context=provider_base.ProviderCallContext(
+                session_id=session_id,
+            ),
         )
 
     assert bodies[0]["prompt_cache_key"] == bodies[1]["prompt_cache_key"]
     assert bodies[0]["prompt_cache_key"] != bodies[2]["prompt_cache_key"]
-    assert all("service_tier" not in body for body in bodies)
 
 
 @pytest.mark.asyncio

tests/providers/test_provider_retry.py

@@ -431,13 +431,17 @@ async def test_image_retry_discards_provider_state_with_images(
 
     response = await provider.chat_with_retry(
         messages=messages,
-        provider_context=ProviderCallContext(conversation_state=state),
+        provider_context=ProviderCallContext(
+            conversation_state=state,
+            session_id="webui:cache-test",
+        ),
     )
 
     assert response.content == "ok, no image"
     retry_context = provider.contexts[-1]
     assert isinstance(retry_context, ProviderCallContext)
     assert retry_context.conversation_state is None
+    assert retry_context.session_id == "webui:cache-test"
     public_content = messages[0]["content"]
     if isinstance(public_content, list):
         assert all(block.get("type") != "image_url" for block in public_content)