fix(llm): resolve score runtime config from saved profiles

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

tests/webapp/test_llm_profiles_api.py

@@ -1,6 +1,7 @@
 """Integration tests for /api/llm-profiles endpoints."""
 import pytest
 from fastapi.testclient import TestClient
+from unittest.mock import patch
 
 
 @pytest.fixture()
@@ -41,19 +42,23 @@ def test_update_profile(client):
     pid = client.post("/api/llm-profiles", json=body).json()["profile_id"]
 
     upd = {"name": "New", "model": "m2", "base_url": "http://x/v1", "api_key": "k", "timeout_seconds": 60}
-    resp = client.put(f"/api/llm-profiles/{pid}", json=upd)
+    with patch("webapp.services.inline_scorer.inline_scorer.invalidate_cache") as invalidate:
+        resp = client.put(f"/api/llm-profiles/{pid}", json=upd)
     assert resp.status_code == 200
     assert resp.json()["name"] == "New"
     assert resp.json()["timeout_seconds"] == 60
+    invalidate.assert_called_once()
 
 
 def test_delete_profile(client):
     body = {"name": "Del", "model": "m", "base_url": "http://x/v1", "api_key": "k"}
     pid = client.post("/api/llm-profiles", json=body).json()["profile_id"]
-    resp = client.delete(f"/api/llm-profiles/{pid}")
+    with patch("webapp.services.inline_scorer.inline_scorer.invalidate_cache") as invalidate:
+        resp = client.delete(f"/api/llm-profiles/{pid}")
     assert resp.status_code == 200
     assert resp.json()["deleted"] is True
     assert len(client.get("/api/llm-profiles").json()["profiles"]) == 0
+    invalidate.assert_called_once()
 
 
 def test_update_nonexistent(client):
@@ -185,7 +190,7 @@ def test_apply_doc_weights_patches_yaml(tmp_path):
 # ---------------------------------------------------------------------------
 # Connectivity test endpoint tests
 # ---------------------------------------------------------------------------
-from unittest.mock import MagicMock, patch
+from unittest.mock import MagicMock
 
 
 def test_probe_connectivity_success(client):

tests/webapp/test_profile_manager.py

@@ -98,3 +98,52 @@ def test_get_nonexistent(tmp_path):
 def test_delete_nonexistent(tmp_path):
     mgr = _make_manager(tmp_path)
     assert mgr.delete("does-not-exist") is False
+
+
+def test_resolve_openai_client_kwargs_prefers_matching_profile(tmp_path, monkeypatch):
+    """Metric runtime should prefer the saved LLM Profile over .env defaults."""
+    from rag_eval.metrics.factory import _resolve_openai_client_kwargs
+    from rag_eval.settings import EvaluationSettings
+    import webapp.services.profile_manager as pm_mod
+
+    mgr = _make_manager(tmp_path)
+    mgr.create(
+        name="Judge",
+        model="gpt-5.5",
+        base_url="http://39.107.88.131:13000",
+        api_key="sk-profile",
+        timeout_seconds=300,
+    )
+    monkeypatch.setattr(pm_mod, "profile_manager", mgr)
+
+    settings = EvaluationSettings(
+        OPENAI_API_KEY="sk-env",
+        OPENAI_BASE_URL="http://env-base/v1",
+        OPENAI_TIMEOUT_SECONDS=30,
+    )
+
+    kwargs = _resolve_openai_client_kwargs("gpt-5.5", settings)
+    assert kwargs["api_key"] == "sk-profile"
+    assert kwargs["base_url"] == "http://39.107.88.131:13000"
+    assert kwargs["timeout"] == 300.0
+
+
+def test_resolve_openai_client_kwargs_falls_back_to_env(tmp_path, monkeypatch):
+    """When no saved profile matches, .env settings remain the fallback."""
+    from rag_eval.metrics.factory import _resolve_openai_client_kwargs
+    from rag_eval.settings import EvaluationSettings
+    import webapp.services.profile_manager as pm_mod
+
+    mgr = _make_manager(tmp_path)
+    monkeypatch.setattr(pm_mod, "profile_manager", mgr)
+
+    settings = EvaluationSettings(
+        OPENAI_API_KEY="sk-env",
+        OPENAI_BASE_URL="http://env-base/v1",
+        OPENAI_TIMEOUT_SECONDS=45,
+    )
+
+    kwargs = _resolve_openai_client_kwargs("gpt-5", settings)
+    assert kwargs["api_key"] == "sk-env"
+    assert kwargs["base_url"] == "http://env-base/v1"
+    assert kwargs["timeout"] == 45.0