evalvault 1.76.0__py3-none-any.whl → 1.77.0__py3-none-any.whl

evalvault/adapters/outbound/phoenix/sync_service.py

@@ -15,7 +15,7 @@ class PhoenixSyncError(RuntimeError):
 def _normalize_base_url(endpoint: str) -> str:
     """Convert OTLP endpoint (…/v1/traces) to Phoenix REST base URL."""
 
-    if not endpoint:
+    if not isinstance(endpoint, str) or not endpoint:
         return "http://localhost:6006"
     base = endpoint.strip()
     suffix = "/v1/traces"

evalvault/adapters/outbound/report/markdown_adapter.py

@@ -50,6 +50,15 @@ class MarkdownReportAdapter:
         # 통계 분석
         if bundle.statistical:
             sections.append(self._generate_statistical_section(bundle.statistical))
+            sections.append(self._generate_reason_section(bundle.statistical))
+            sections.append(self._generate_meaning_section(bundle.statistical))
+            sections.append(self._generate_dataset_delta_section(bundle.statistical))
+            sections.append(self._generate_improvement_plan_section(bundle.statistical))
+        else:
+            sections.append(self._generate_reason_section(None))
+            sections.append(self._generate_meaning_section(None))
+            sections.append(self._generate_dataset_delta_section(None))
+            sections.append(self._generate_improvement_plan_section(None))
 
         # NLP 분석
         if include_nlp and bundle.has_nlp and bundle.nlp:
@@ -208,6 +217,89 @@ class MarkdownReportAdapter:
 
         return "\n".join(lines)
 
+    def _generate_reason_section(self, stat: StatisticalAnalysis | None) -> str:
+        lines = ["## 원인/근거"]
+        if stat is None:
+            lines.append(
+                "- 통계 분석 결과가 없어 원인/근거를 도출할 수 없습니다. (추가 데이터 필요)"
+            )
+            return "\n".join(lines)
+
+        if stat.low_performers:
+            for lp in stat.low_performers[:5]:
+                lines.append(
+                    f"- {lp.test_case_id}: {lp.metric_name} {lp.score:.2f} < {lp.threshold:.2f}"
+                )
+        elif stat.insights:
+            for insight in stat.insights[:5]:
+                lines.append(f"- {insight}")
+        else:
+            lines.append("- 추가 데이터 필요")
+        return "\n".join(lines)
+
+    def _generate_meaning_section(self, stat: StatisticalAnalysis | None) -> str:
+        lines = ["## 결과 의미"]
+        if stat is None:
+            lines.append("- 통계 분석 결과가 없어 의미를 해석할 수 없습니다. (추가 데이터 필요)")
+            return "\n".join(lines)
+
+        if stat.overall_pass_rate < 0.7:
+            lines.append("- 전체 통과율이 낮아 사용자 신뢰/정확성 리스크가 큽니다.")
+        else:
+            lines.append("- 전체 통과율이 기준 이상으로 기본 품질은 유지됩니다.")
+
+        low_metrics = [
+            metric for metric, rate in (stat.metric_pass_rates or {}).items() if rate < 0.7
+        ]
+        if low_metrics:
+            metrics_str = ", ".join(sorted(low_metrics)[:6])
+            lines.append(f"- 기준 미달 메트릭: {metrics_str}")
+        return "\n".join(lines)
+
+    def _generate_dataset_delta_section(self, stat: StatisticalAnalysis | None) -> str:
+        lines = ["## 데이터셋 차이"]
+        if stat is None:
+            lines.append("- 데이터셋 기준 차이를 판단할 수 없습니다. (추가 데이터 필요)")
+            return "\n".join(lines)
+
+        low_metrics = [
+            metric for metric, rate in (stat.metric_pass_rates or {}).items() if rate < 0.7
+        ]
+        if low_metrics:
+            lines.append("- 데이터셋 기준 미달 지표: " + ", ".join(sorted(low_metrics)[:6]))
+        else:
+            lines.append("- 데이터셋 기준 미달 지표가 없습니다.")
+        return "\n".join(lines)
+
+    def _generate_improvement_plan_section(self, stat: StatisticalAnalysis | None) -> str:
+        lines = ["## 개선 방향"]
+        if stat is None:
+            lines.append("- 개선 방향 도출을 위한 분석 결과가 부족합니다. (추가 데이터 필요)")
+            return "\n".join(lines)
+
+        action_map = {
+            "context_precision": "랭커/리랭커 도입 및 상위 문서 필터링 강화",
+            "context_recall": "검색 범위 확장 또는 하드 네거티브 추가",
+            "mrr": "상위 K 재정렬 및 쿼리 재작성 적용",
+            "ndcg": "랭킹 품질 지표 최적화(리랭킹/하이브리드 검색)",
+            "hit_rate": "검색 후보군 확대 또는 인덱싱 개선",
+            "answer_relevancy": "답변 포맷/질문 의도 정렬 프롬프트 강화",
+            "faithfulness": "근거 인용/검증 단계 추가",
+            "factual_correctness": "정답 검증 규칙 강화 및 근거 필터링",
+            "semantic_similarity": "정답 기준 문장 재정의 및 평가셋 보강",
+        }
+
+        low_metrics = [
+            metric for metric, rate in (stat.metric_pass_rates or {}).items() if rate < 0.7
+        ]
+        if low_metrics:
+            for metric in sorted(low_metrics)[:5]:
+                action = action_map.get(metric, "실험을 통해 개선 방향을 재검증")
+                lines.append(f"- {metric}: {action}")
+        else:
+            lines.append("- 개선 대상 지표가 명확하지 않습니다. (추가 데이터 필요)")
+        return "\n".join(lines)
+
     def _generate_nlp_section(self, nlp: NLPAnalysis) -> str:
         """NLP 분석 섹션 생성."""
         lines = ["## NLP 분석"]

evalvault/adapters/outbound/storage/factory.py

@@ -19,12 +19,9 @@ def build_storage_adapter(
 ) -> StoragePort:
     resolved_settings = settings or Settings()
 
-    if db_path is not None:
-        return SQLiteStorageAdapter(db_path=db_path)
-
     backend = getattr(resolved_settings, "db_backend", "postgres")
     if backend == "sqlite":
-        resolved_db_path = resolved_settings.evalvault_db_path
+        resolved_db_path = db_path or resolved_settings.evalvault_db_path
         if resolved_db_path is None:
             raise RuntimeError("SQLite backend selected but evalvault_db_path is not set.")
         return SQLiteStorageAdapter(db_path=resolved_db_path)

evalvault/adapters/outbound/tracker/phoenix_adapter.py

@@ -164,6 +164,20 @@ class PhoenixAdapter(TrackerPort):
             from opentelemetry.trace import format_span_id
 
             span_id = format_span_id(span.get_span_context().span_id)
+            spans_client = getattr(client, "spans", None)
+            add_span_annotation = (
+                getattr(spans_client, "add_span_annotation", None) if spans_client else None
+            )
+            if callable(add_span_annotation):
+                add_span_annotation(
+                    annotation_name=name,
+                    annotator_kind="CODE",
+                    span_id=span_id,
+                    label=label,
+                    score=score,
+                    explanation=explanation,
+                )
+                return
             client.annotations.add_span_annotation(
                 annotation_name=name,
                 annotator_kind="CODE",
@@ -339,6 +353,7 @@ class PhoenixAdapter(TrackerPort):
 
         # Calculate per-metric summary
         metric_summary = {}
+        total_count = len(run.results) if run.results else 0
         for metric_name in run.metrics_evaluated:
             passed_count = sum(
                 1
@@ -351,9 +366,9 @@ class PhoenixAdapter(TrackerPort):
                 "average_score": round(avg_score, 4) if avg_score else 0.0,
                 "threshold": threshold,
                 "passed": passed_count,
-                "failed": len(run.results) - passed_count,
-                "total": len(run.results),
-                "pass_rate": round(passed_count / len(run.results), 4) if run.results else 0.0,
+                "failed": total_count - passed_count,
+                "total": total_count,
+                "pass_rate": round(passed_count / total_count, 4) if total_count else 0.0,
             }
 
         # Start root trace
@@ -805,11 +820,13 @@ class PhoenixAdapter(TrackerPort):
         if not output_preview and data.generation and data.generation.response:
             output_preview = sanitize_text(data.generation.response, max_chars=MAX_LOG_CHARS)
         if not output_preview and data.retrieval:
-            previews = [
-                sanitize_text(doc.content, max_chars=MAX_CONTEXT_CHARS)
-                for doc in data.retrieval.candidates
-                if doc.content
-            ]
+            previews: list[str] = []
+            for doc in data.retrieval.candidates:
+                if not doc.content:
+                    continue
+                preview = sanitize_text(doc.content, max_chars=MAX_CONTEXT_CHARS)
+                if preview:
+                    previews.append(preview)
             output_preview = "\n".join(previews[:3])
         if output_preview:
             span.set_attribute("rag.final_answer", output_preview)

evalvault/config/runtime_services.py

@@ -0,0 +1,122 @@
+from __future__ import annotations
+
+import logging
+import socket
+import subprocess
+from dataclasses import dataclass
+from shutil import which
+from urllib.parse import urlparse
+
+from evalvault.config.settings import Settings, is_production_profile
+
+logger = logging.getLogger(__name__)
+
+_PHOENIX_CONTAINER = "evalvault-phoenix"
+
+
+@dataclass(frozen=True)
+class Endpoint:
+    host: str
+    port: int
+
+
+def _is_local_host(host: str | None) -> bool:
+    if not host:
+        return False
+    return host in {"localhost", "127.0.0.1", "0.0.0.0"}
+
+
+def _port_is_open(host: str, port: int) -> bool:
+    try:
+        with socket.create_connection((host, port), timeout=0.2):
+            return True
+    except OSError:
+        return False
+
+
+def _parse_http_endpoint(url: str | None, default_port: int) -> Endpoint | None:
+    if not url or not isinstance(url, str):
+        return None
+    parsed = urlparse(url)
+    if parsed.scheme and parsed.scheme not in {"http", "https"}:
+        return None
+    host = parsed.hostname or ""
+    port = parsed.port or default_port
+    if not host or port <= 0:
+        return None
+    return Endpoint(host=host, port=port)
+
+
+def _start_mlflow(port: int) -> bool:
+    if which("mlflow") is None:
+        logger.warning("MLflow CLI not found. Install with: uv sync --extra mlflow")
+        return False
+    try:
+        subprocess.Popen(
+            ["mlflow", "server", "--host", "0.0.0.0", "--port", str(port)],
+            stdout=subprocess.DEVNULL,
+            stderr=subprocess.DEVNULL,
+        )
+        logger.info("Started MLflow server on port %s", port)
+        return True
+    except Exception as exc:  # pragma: no cover - safety net
+        logger.warning("Failed to start MLflow server: %s", exc)
+        return False
+
+
+def _start_phoenix(port: int) -> bool:
+    if which("docker") is None:
+        logger.warning("Docker not found. Phoenix auto-start skipped.")
+        return False
+    try:
+        start = subprocess.run(
+            ["docker", "start", _PHOENIX_CONTAINER],
+            check=False,
+            capture_output=True,
+            text=True,
+        )
+        if start.returncode != 0:
+            subprocess.run(
+                [
+                    "docker",
+                    "run",
+                    "-d",
+                    "-p",
+                    f"{port}:6006",
+                    "--name",
+                    _PHOENIX_CONTAINER,
+                    "arizephoenix/phoenix:latest",
+                ],
+                check=False,
+                capture_output=True,
+                text=True,
+            )
+        logger.info("Ensured Phoenix container is running on port %s", port)
+        return True
+    except Exception as exc:  # pragma: no cover - safety net
+        logger.warning("Failed to start Phoenix container: %s", exc)
+        return False
+
+
+def ensure_local_observability(settings: Settings) -> None:
+    if is_production_profile(settings.evalvault_profile):
+        return
+
+    phoenix_endpoint = _parse_http_endpoint(
+        getattr(settings, "phoenix_endpoint", None) or "http://localhost:6006/v1/traces",
+        6006,
+    )
+    if (
+        phoenix_endpoint
+        and _is_local_host(phoenix_endpoint.host)
+        and not _port_is_open(phoenix_endpoint.host, phoenix_endpoint.port)
+    ):
+        _start_phoenix(phoenix_endpoint.port)
+
+    mlflow_endpoint = _parse_http_endpoint(getattr(settings, "mlflow_tracking_uri", None), 5000)
+    if (
+        mlflow_endpoint
+        and _is_local_host(mlflow_endpoint.host)
+        and not _port_is_open(mlflow_endpoint.host, mlflow_endpoint.port)
+    ):
+        _start_mlflow(mlflow_endpoint.port)

{evalvault-1.76.0.dist-info → evalvault-1.77.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: evalvault
-Version: 1.76.0
+Version: 1.77.0
 Summary: RAG evaluation system using Ragas with Phoenix/Langfuse tracing
 Project-URL: Homepage, https://github.com/ntts9990/EvalVault
 Project-URL: Documentation, https://github.com/ntts9990/EvalVault#readme
@@ -66,6 +66,7 @@ Requires-Dist: ijson>=3.3.0; extra == 'dev'
 Requires-Dist: kiwipiepy>=0.18.0; extra == 'dev'
 Requires-Dist: langchain-anthropic; extra == 'dev'
 Requires-Dist: lm-eval[api]>=0.4.0; extra == 'dev'
+Requires-Dist: manim>=0.18.0; extra == 'dev'
 Requires-Dist: mkdocs-material>=9.5.0; extra == 'dev'
 Requires-Dist: mkdocs>=1.5.0; extra == 'dev'
 Requires-Dist: mkdocstrings[python]>=0.24.0; extra == 'dev'

{evalvault-1.76.0.dist-info → evalvault-1.77.0.dist-info}/RECORD

@@ -5,8 +5,8 @@ evalvault/mkdocs_helpers.py,sha256=1AKVQ1W2_VO4qclhfyefyU9Dz1Hzkh1DWDwsFMe24jc,3
 evalvault/adapters/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 evalvault/adapters/inbound/__init__.py,sha256=SG1svel1PwqetnqVpKFLSv612_WwGwLTbFpYgwk6FMw,166
 evalvault/adapters/inbound/api/__init__.py,sha256=LeVVttCA3tLKoHA2PO4z3y8VkfVcf3Bq8CZSzo91lf4,34
-evalvault/adapters/inbound/api/adapter.py,sha256=wzudfkMALcX3IH7MfG-dYc5j-nzxBdDBT1X8_vJWick,86848
-evalvault/adapters/inbound/api/main.py,sha256=QgLxzHEy7aycGKIFLtN12tWTjnpWLtQ2XDXKV_2FDvg,7531
+evalvault/adapters/inbound/api/adapter.py,sha256=fYI8NBg4Ropav5W4sVwSus6SFAkhXdWoAEY70_UryaY,87797
+evalvault/adapters/inbound/api/main.py,sha256=B22zDra7MawI_fbQ4uJOzlf7Lrbxy7hycebT9kO9pL0,7651
 evalvault/adapters/inbound/api/routers/__init__.py,sha256=q07_YF9TnBl68bqcRCvhPU4-zRTyvmPoHVehwO6W7QM,19
 evalvault/adapters/inbound/api/routers/benchmark.py,sha256=yevntbZcNtMvbVODsITUBgR1Ka4pdFQrXBJJ4K4Jyr4,4477
 evalvault/adapters/inbound/api/routers/calibration.py,sha256=FzEOe7KqDtfGDlOKREsAoYId1jTILyqkDm_Y3LyHk54,4130
@@ -18,10 +18,10 @@ evalvault/adapters/inbound/api/routers/mcp.py,sha256=yHANV7qIXig-7YSiQgXzSTuabqF
 evalvault/adapters/inbound/api/routers/pipeline.py,sha256=VCoPYg69fg2hc5Z8oWIPbHzYYT2NnGoRZW48C48lApE,17442
 evalvault/adapters/inbound/api/routers/runs.py,sha256=eQu_6Vp-8lcp4oVaEFhlmR8uCyQ8ys_-VRHO5AKDQvs,43617
 evalvault/adapters/inbound/cli/__init__.py,sha256=a42flC5NK-VfbdbBrE49IrUL5zAyKdXZYJVM6E3NTE0,675
-evalvault/adapters/inbound/cli/app.py,sha256=ytNgHRg9ZTAl33AkB1wIL8RKfQ_Cf8fsy0gSsLTs7Ew,1603
+evalvault/adapters/inbound/cli/app.py,sha256=7owV7vDZCZlAByHnalY84mmPjxdf1nSciigTJ_smr20,1774
 evalvault/adapters/inbound/cli/commands/__init__.py,sha256=TB3evGuv3_AQoudWiR2bH5EH_AY9f9E7lQcSOWH6TO8,4091
 evalvault/adapters/inbound/cli/commands/agent.py,sha256=YlOYMEzzS1aSKDKD_a7UK3St18X6GXGkdTatrzyd8Zc,7555
-evalvault/adapters/inbound/cli/commands/analyze.py,sha256=kSnYGx54v6wWNwWJYqZYBXMR0OR6v8YDxyfpF3eveSc,50565
+evalvault/adapters/inbound/cli/commands/analyze.py,sha256=p7J4bcn9tp2YR5v8655z_xiqelx8Y-sxD4rHBTvasdU,50745
 evalvault/adapters/inbound/cli/commands/api.py,sha256=YdbJ_-QEajnFcjTa7P2heLMjFKpeQ4nWP_p-HvfYkEo,1943
 evalvault/adapters/inbound/cli/commands/artifacts.py,sha256=bE8FQxmnU0mMIAPx5en8aKrtfNNkrbWoLxIX4ZT9D5c,3776
 evalvault/adapters/inbound/cli/commands/benchmark.py,sha256=rcmLcId5w_cM5ZfMzu2V49B-pDAZMXBcYgiB8v1X0Eo,41306
@@ -39,15 +39,15 @@ evalvault/adapters/inbound/cli/commands/history.py,sha256=dGkoRz59FZCbs-Af29MxgV
 evalvault/adapters/inbound/cli/commands/init.py,sha256=7q86fUeBVA08fU_N0lAV6Lakxirq4val2jIyALlDy3E,8822
 evalvault/adapters/inbound/cli/commands/kg.py,sha256=ycV9Xj6SUUJLTyTfLZcjXDVLcZqwo7Gw878ZhZAeDoc,19155
 evalvault/adapters/inbound/cli/commands/langfuse.py,sha256=aExhZ5WYT0FzJI4v1sF-a1jqy9b1BF46_HBtfiQjVGI,4085
-evalvault/adapters/inbound/cli/commands/method.py,sha256=2DuMD1AMaQr9gl6xAQty_YKdq5MA9Y-FMYMmERnf9-4,19090
+evalvault/adapters/inbound/cli/commands/method.py,sha256=Bh-kcSEk2QFeLLUrL9OB8baHndiR4FmpMoolDEDF4fM,19100
 evalvault/adapters/inbound/cli/commands/ops.py,sha256=jh0ryYIDvj2VDhOR6bESPHnyogtzAfSs81gKMGlOczs,4042
 evalvault/adapters/inbound/cli/commands/phoenix.py,sha256=LQi3KTLq1ybjjBuz92oQ6lYyBS3mHrCHk0qe-7bqB4U,15611
 evalvault/adapters/inbound/cli/commands/pipeline.py,sha256=nBsgSG9RvwDRy_X-01DqZe0jS9zpp1j4hkccni3sruc,12253
 evalvault/adapters/inbound/cli/commands/profile_difficulty.py,sha256=PF8MbYzufbNVlKKnBp6JF9PrBe-ncGDOe8JHWW0L278,6329
 evalvault/adapters/inbound/cli/commands/prompts.py,sha256=ExbTG_8yZupuw6em6OPOdTOMc4T-YPc7tUx5_cbXeiU,26708
 evalvault/adapters/inbound/cli/commands/regress.py,sha256=VzwTucxJav71Xmd18qCb93XG_mTvfjmW52SQRv8Sgu4,19082
-evalvault/adapters/inbound/cli/commands/run.py,sha256=H65pQ_bbUST_d6_OdUDU9ZS-FMLE8FMy4KXLWj11Qzg,136577
-evalvault/adapters/inbound/cli/commands/run_helpers.py,sha256=yFt7jzZX8NEClCJdwOt9LfrpIhD5J_m9HD-XSrSLP0I,47994
+evalvault/adapters/inbound/cli/commands/run.py,sha256=G_6AhvD-VByY1fj3JqYvmrYUleeyAdXRLY3a-rG7IIE,136860
+evalvault/adapters/inbound/cli/commands/run_helpers.py,sha256=VWZKtkY4MwMpLbgq2zzl3uSu6OrXcixvpn3GYgBgBtM,48055
 evalvault/adapters/inbound/cli/commands/stage.py,sha256=XiPfJrdtUWrdSw5EzLgvz5WEEBW1qz1FALEhDYamMjE,23236
 evalvault/adapters/inbound/cli/utils/__init__.py,sha256=QPNKneZS-Z-tTnYYxtgJXgcJWY6puUlRQcKrn7Mlv1M,685
 evalvault/adapters/inbound/cli/utils/analysis_io.py,sha256=RHkKEq4e-PtbtRDlXAJWU80RYHNPw-O5V9_GujdaGfc,13393
@@ -84,7 +84,7 @@ evalvault/adapters/outbound/analysis/embedding_searcher_module.py,sha256=DcAMuV7
 evalvault/adapters/outbound/analysis/hybrid_rrf_module.py,sha256=kaHSc7z3Jg_KrRLBqPMTV_9XXsL6v1dmbz-3dDO6IMw,3255
 evalvault/adapters/outbound/analysis/hybrid_weighted_module.py,sha256=AO-7thmnFGerUDWd8l9ydxeAkHkACo7Raf9O0RfW_nE,3671
 evalvault/adapters/outbound/analysis/hypothesis_generator_module.py,sha256=tx9fWgS0rBoK5eJPmwK5POoV78yN03hkFmWhCx71Ln0,13337
-evalvault/adapters/outbound/analysis/llm_report_module.py,sha256=ymYtSnux1NYK_P9dt4KLh6dL60KjtPBluClwFWRakwU,70455
+evalvault/adapters/outbound/analysis/llm_report_module.py,sha256=sYwgeS2W3jJRJzvFMYb3tnXag9x__iAVVP-8YzHXACU,91812
 evalvault/adapters/outbound/analysis/low_performer_extractor_module.py,sha256=Pt0Tmtc5Etqp_3SBDCPAzqWI2EF9woSg0mmBucEHlQw,1291
 evalvault/adapters/outbound/analysis/model_analyzer_module.py,sha256=28rHdXBXYIFpLHixbbZcv6-j2QVgl3yaGN0vU1Q0gFc,2682
 evalvault/adapters/outbound/analysis/morpheme_analyzer_module.py,sha256=Hrh4mluMsOhQHPrliD2w0FVKokJpfikXOFKT6sNwk74,4158
@@ -186,12 +186,12 @@ evalvault/adapters/outbound/nlp/korean/toolkit.py,sha256=iMaY5VueC5P0rxm5SCdz69C
 evalvault/adapters/outbound/nlp/korean/toolkit_factory.py,sha256=x3v-AAkVInOabC4PtOtStsZrFnHun0IOqZDyQGaQVm8,586
 evalvault/adapters/outbound/ops/__init__.py,sha256=_QiDVPuiYWkIwW_ELEVKD_v6dLojjyvIJWs4qVNxehw,164
 evalvault/adapters/outbound/ops/report_renderer.py,sha256=mezVKdIsnJSNvBW6xkhpNG3MOFXHZLZspmHk5o-e8Cg,6354
-evalvault/adapters/outbound/phoenix/sync_service.py,sha256=3rNtolFzwbRbmsP7fCh6lMDcpEf428mLqpZuw1tYW3M,13271
+evalvault/adapters/outbound/phoenix/sync_service.py,sha256=D5OUfArTq6K9D7oCe3CbIKOqqfRK9r4WoyqFcL03uxs,13304
 evalvault/adapters/outbound/report/__init__.py,sha256=8VeMrfj63mDR-xUHct-drNNBA5M-m-B7sgC1qUJF7g4,660
 evalvault/adapters/outbound/report/ci_report_formatter.py,sha256=5YD8BwtOjLnHcNbbG0HJziOifD9BDhBtZT1oItd6zJE,1233
 evalvault/adapters/outbound/report/dashboard_generator.py,sha256=g0SANOrOS_mSfLfKbfbxflxZmCNbZGrGzA0g03_Yb48,8356
 evalvault/adapters/outbound/report/llm_report_generator.py,sha256=i_iXfY8qutIb8TsvLKyMLnijsA0yiNJ3rBEFg4zVqcE,26858
-evalvault/adapters/outbound/report/markdown_adapter.py,sha256=5PS72h_qe4ZtYs-umhX5TqQL2k5SuDaCUc6rRw9AKRw,16761
+evalvault/adapters/outbound/report/markdown_adapter.py,sha256=5KaMeStOLgKsBDTf2DeDbbnpN_eWp2PhkdAwpe1KrLo,21346
 evalvault/adapters/outbound/report/pr_comment_formatter.py,sha256=FxWWfZQU5ErejVqE_F8rKoUxoBTbbmw_ok9xbEYiA4E,1661
 evalvault/adapters/outbound/retriever/__init__.py,sha256=o5qK4pIrlXm0mIs_fo0aLPQHuHkrmrXKc6ZSg8a0t4g,201
 evalvault/adapters/outbound/retriever/graph_rag_adapter.py,sha256=xTI7uMFp4WKstg1s2zY9R_QsWPA-Rz_KUnzekwso8z0,11790
@@ -199,7 +199,7 @@ evalvault/adapters/outbound/retriever/pgvector_store.py,sha256=gf2oO_aWpkaDJxBhc
 evalvault/adapters/outbound/storage/__init__.py,sha256=n5R6thAPTx1leSwv6od6nBWcLWFa-UYD6cOLzN89T8I,614
 evalvault/adapters/outbound/storage/base_sql.py,sha256=WzA57p11m65zf-F7jjiJ9GG0BxnnAahKRckVOiK3xcY,58441
 evalvault/adapters/outbound/storage/benchmark_storage_adapter.py,sha256=Qgf9xSSIkYQRpG4uLzcUdoYO9LTQDQ4tFRkkMYer-WA,9803
-evalvault/adapters/outbound/storage/factory.py,sha256=PVtU4C1NGdAnghMgs4k_p36KqNMJUQjBFVwg9W4MBxw,1980
+evalvault/adapters/outbound/storage/factory.py,sha256=lU5nsBbTPtt_yFbw4GHms2lbdY7AACesWE2D8iK6oXU,1909
 evalvault/adapters/outbound/storage/postgres_adapter.py,sha256=SZb4Dx2ZYNu-pDqX4rVgT4bRfJ2cQ75DXcWIIEu8aws,59933
 evalvault/adapters/outbound/storage/postgres_schema.sql,sha256=CIQn1S93ZrH1B_qumePf5aKi-25L284QmvpllnUfeEI,12375
 evalvault/adapters/outbound/storage/schema.sql,sha256=LEtrKFpa1SbIBN-igkNiQXJqVLGp-liX-6KR_sZAMwM,14283
@@ -214,7 +214,7 @@ evalvault/adapters/outbound/tracker/__init__.py,sha256=Suu5BznOK5uTuD5_jS8JMZd8R
 evalvault/adapters/outbound/tracker/langfuse_adapter.py,sha256=uI-t5v9AC5VUMYsIc1FHYImourZeErGMXB0_prOMErc,18839
 evalvault/adapters/outbound/tracker/log_sanitizer.py,sha256=ilKTTSzsHslQYc-elnWu0Z3HKNNw1D1iI0_cCvYbo1M,2653
 evalvault/adapters/outbound/tracker/mlflow_adapter.py,sha256=8vo53d2ZpcYDz3J798iru7Sk-Wm3XhARJMVmS5d6pRE,13495
-evalvault/adapters/outbound/tracker/phoenix_adapter.py,sha256=EXEhoOFTFNSVuSnN2wFIH3H1kq4T14601sGIcRIy8hs,32947
+evalvault/adapters/outbound/tracker/phoenix_adapter.py,sha256=ajH8B9kgmUBywrko20Fo8DUi-kbzsTiNZZUZSwfBfOA,33663
 evalvault/config/__init__.py,sha256=UCgeDx62M2gOuFvdN29wWwny2fdH4bPY_uUC3-42eDw,1297
 evalvault/config/agent_types.py,sha256=EP2Pv3ZtOzDXIvIa-Hnd1to9JIbMUtGitrlwzZtx0Ys,13418
 evalvault/config/domain_config.py,sha256=rOgNA2T8NWlDzcEFC0shdUCCww0lI1E5fUm5QrKQSZI,9264
@@ -222,6 +222,7 @@ evalvault/config/instrumentation.py,sha256=BW2a3OqHH6aQNjjlL0oyEI6K3pXuORy6x__-Z
 evalvault/config/langfuse_support.py,sha256=DEzVMfMGGf1V45W_2oUG-NCDfsYI4UUdnYJIgBSrN2o,582
 evalvault/config/model_config.py,sha256=KlzDbGyDLeOGE7ElekFFk5YjjT5u8i6KO2B4EyZkLnI,3542
 evalvault/config/phoenix_support.py,sha256=w2W8UQTprac8BWdIfV9kPjr2sEj7EF9pNBXX2JB1x4M,13810
+evalvault/config/runtime_services.py,sha256=-dktvBVmqudfk0Q07A1vY8KuhtJTnoU9ZxLGR5_ifmw,3642
 evalvault/config/secret_manager.py,sha256=YjPMuNqeBrAR2BzCJvsBNUExaU4TBSFyZ8kVYZZifqA,4172
 evalvault/config/settings.py,sha256=2EoO6WutkwD2F7rkyECxx5-VenivntOXPZwemVQ9ZtU,20696
 evalvault/config/playbooks/improvement_playbook.yaml,sha256=9F9WVVCydFfz6zUuGYzZ4PKdW1LLtcBKVF36T7xT764,26965
@@ -364,8 +365,8 @@ evalvault/reports/__init__.py,sha256=Bb1X4871msAN8I6PM6nKGED3psPwZt88hXZBAOdH06Y
 evalvault/reports/release_notes.py,sha256=pZj0PBFT-4F_Ty-Kv5P69BuoOnmTCn4kznDcORFJd0w,4011
 evalvault/scripts/__init__.py,sha256=NwEeIFQbkX4ml2R_PhtIoNtArDSX_suuoymgG_7Kwso,89
 evalvault/scripts/regression_runner.py,sha256=SxZori5BZ8jVQ057Mf5V5FPgIVDccrV5oRONmnhuk8w,8438
-evalvault-1.76.0.dist-info/METADATA,sha256=HleQUSCdC3UNzeCVN4PdvcOG5Jx-dYfMgaq8VwRnlYQ,8752
-evalvault-1.76.0.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
-evalvault-1.76.0.dist-info/entry_points.txt,sha256=Oj9Xc5gYcyUYYNmQfWI8NYGw7nN-3M-h2ipHIMlVn6o,65
-evalvault-1.76.0.dist-info/licenses/LICENSE.md,sha256=3RNWY4jjtrQ_yYa-D-7I3XO12Ti7YzxsLV_dpykujvo,11358
-evalvault-1.76.0.dist-info/RECORD,,
+evalvault-1.77.0.dist-info/METADATA,sha256=5ALEi-ipXLZDaMenmFQFAlYOsmHjNHtUWuR_yxzCpIM,8797
+evalvault-1.77.0.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
+evalvault-1.77.0.dist-info/entry_points.txt,sha256=Oj9Xc5gYcyUYYNmQfWI8NYGw7nN-3M-h2ipHIMlVn6o,65
+evalvault-1.77.0.dist-info/licenses/LICENSE.md,sha256=3RNWY4jjtrQ_yYa-D-7I3XO12Ti7YzxsLV_dpykujvo,11358
+evalvault-1.77.0.dist-info/RECORD,,