feat(platform): close AI expense value loop

Add tenant-safe value, telemetry, connector, commercial, and production-readiness foundations.

server/tests/test_runtime_chat_service.py

@@ -39,7 +39,9 @@ def test_runtime_chat_fails_over_to_backup_before_retrying_main(monkeypatch) ->
                 "apiKey": "secret",
             }
 
-        def fake_request_chat_completion(config, messages, *, max_tokens, temperature, timeout_seconds):
+        def fake_request_chat_completion(
+            config, messages, *, max_tokens, temperature, timeout_seconds
+        ):
             del messages, max_tokens, temperature, timeout_seconds
             calls.append(config["slot"])
             if config["slot"] == "main":
@@ -70,7 +72,9 @@ def test_runtime_chat_complete_with_trace_records_slot_failover(monkeypatch) ->
                 "apiKey": "secret",
             }
 
-        def fake_request_chat_completion(config, messages, *, max_tokens, temperature, timeout_seconds):
+        def fake_request_chat_completion(
+            config, messages, *, max_tokens, temperature, timeout_seconds
+        ):
             del messages, max_tokens, temperature, timeout_seconds
             if config["slot"] == "main":
                 raise RuntimeError("incorrect api key")
@@ -104,7 +108,9 @@ def test_runtime_chat_does_not_rehit_failed_slots_during_cooldown(monkeypatch) -
                 "apiKey": "secret",
             }
 
-        def fake_request_chat_completion(config, messages, *, max_tokens, temperature, timeout_seconds):
+        def fake_request_chat_completion(
+            config, messages, *, max_tokens, temperature, timeout_seconds
+        ):
             del messages, max_tokens, temperature, timeout_seconds
             calls.append(config["slot"])
             raise RuntimeError("unavailable")
@@ -134,7 +140,7 @@ def test_runtime_chat_disables_glm_thinking_for_direct_user_answers(monkeypatch)
 
         monkeypatch.setattr("app.services.runtime_chat._send_json_request", fake_send_json_request)
 
-        answer = service._request_openai_compatible(
+        provider_response = service._request_openai_compatible(
             provider="GLM",
             endpoint="https://open.bigmodel.cn/api/paas/v4/",
             model="glm-5.1",
@@ -145,7 +151,7 @@ def test_runtime_chat_disables_glm_thinking_for_direct_user_answers(monkeypatch)
             timeout_seconds=17,
         )
 
-        assert answer == "ok"
+        assert provider_response.output == "ok"
         assert captured["payload"]["thinking"] == {"type": "disabled"}
         assert captured["timeout_seconds"] == 17
 
@@ -184,7 +190,7 @@ def test_runtime_chat_openai_compatible_tool_call_payload(monkeypatch) -> None:
 
         monkeypatch.setattr("app.services.runtime_chat._send_json_request", fake_send_json_request)
 
-        tool_call = service._request_openai_compatible_tool_call(
+        provider_response = service._request_openai_compatible_tool_call(
             provider="OpenAI Compatible",
             endpoint="https://api.example.com/v1",
             model="gpt-test",
@@ -197,12 +203,16 @@ def test_runtime_chat_openai_compatible_tool_call_payload(monkeypatch) -> None:
             timeout_seconds=19,
         )
 
+        tool_call = provider_response.output
         assert tool_call is not None
         assert tool_call.name == "submit_steward_intent_plan"
         assert tool_call.arguments == {"tasks": []}
         assert captured["url"] == "https://api.example.com/v1/chat/completions"
         assert captured["payload"]["tools"][0]["function"]["name"] == "submit_steward_intent_plan"
-        assert captured["payload"]["tool_choice"]["function"]["name"] == "submit_steward_intent_plan"
+        assert (
+            captured["payload"]["tool_choice"]["function"]["name"]
+            == "submit_steward_intent_plan"
+        )
         assert captured["headers"]["Authorization"] == "Bearer secret"
 
 
@@ -222,7 +232,9 @@ def test_runtime_chat_supports_single_pass_fast_failover(monkeypatch) -> None:
                 "apiKey": "secret",
             }
 
-        def fake_request_chat_completion(config, messages, *, max_tokens, temperature, timeout_seconds):
+        def fake_request_chat_completion(
+            config, messages, *, max_tokens, temperature, timeout_seconds
+        ):
             del messages, max_tokens, temperature
             calls.append((config["slot"], timeout_seconds))
             raise RuntimeError("unavailable")
@@ -242,7 +254,9 @@ def test_runtime_chat_supports_single_pass_fast_failover(monkeypatch) -> None:
         assert calls == [("main", 8), ("backup", 20)]
 
 
-def test_runtime_chat_complete_with_tool_call_fails_over_to_backup_before_retrying_main(monkeypatch) -> None:
+def test_runtime_chat_tool_call_fails_over_to_backup_before_retrying_main(
+    monkeypatch,
+) -> None:
     _clear_runtime_chat_cooldown()
     session_factory = build_session_factory()
     with session_factory() as db:
@@ -258,7 +272,16 @@ def test_runtime_chat_complete_with_tool_call_fails_over_to_backup_before_retryi
                 "apiKey": "secret",
             }
 
-        def fake_request_chat_tool_call(config, messages, *, tools, tool_choice, max_tokens, temperature, timeout_seconds):
+        def fake_request_chat_tool_call(
+            config,
+            messages,
+            *,
+            tools,
+            tool_choice,
+            max_tokens,
+            temperature,
+            timeout_seconds,
+        ):
             del messages, tools, tool_choice, max_tokens, temperature, timeout_seconds
             calls.append(config["slot"])
             if config["slot"] == "main":
@@ -302,7 +325,9 @@ def test_runtime_chat_skips_slot_during_cooldown(monkeypatch) -> None:
                 "apiKey": "secret",
             }
 
-        def fake_request_chat_completion(config, messages, *, max_tokens, temperature, timeout_seconds):
+        def fake_request_chat_completion(
+            config, messages, *, max_tokens, temperature, timeout_seconds
+        ):
             del messages, max_tokens, temperature, timeout_seconds
             calls.append(config["slot"])
             if config["slot"] == "main":
@@ -312,8 +337,14 @@ def test_runtime_chat_skips_slot_during_cooldown(monkeypatch) -> None:
         monkeypatch.setattr(service, "_load_chat_slot", fake_load_chat_slot)
         monkeypatch.setattr(service, "_request_chat_completion", fake_request_chat_completion)
 
-        assert service.complete([{"role": "user", "content": "hello"}], max_attempts=1) == "backup answer"
-        assert service.complete([{"role": "user", "content": "hello again"}], max_attempts=1) == "backup answer"
+        first = service.complete(
+            [{"role": "user", "content": "hello"}], max_attempts=1
+        )
+        second = service.complete(
+            [{"role": "user", "content": "hello again"}], max_attempts=1
+        )
+        assert first == "backup answer"
+        assert second == "backup answer"
         assert calls == ["main", "backup", "backup"]