evalvault 1.73.2__tar.gz → 1.74.0__tar.gz

{evalvault-1.73.2 → evalvault-1.74.0}/.env.offline.example

@@ -14,6 +14,7 @@ EVALVAULT_PROFILE=dev
 # ================================================
 # PostgreSQL (core stack)
 # ================================================
+POSTGRES_IMAGE=postgres:16.4-alpine
 POSTGRES_USER=evalvault
 POSTGRES_PASSWORD=evalvault
 POSTGRES_DB=evalvault
@@ -29,6 +30,14 @@ EVALVAULT_MEMORY_DB_PATH=data/db/evalvault_memory.db
 # ================================================
 CORS_ORIGINS=http://localhost:5173,http://127.0.0.1:5173
 
+# ================================================
+# Docker Base Images (offline builds)
+# ================================================
+EVALVAULT_PYTHON_IMAGE=python:3.12.6-slim
+EVALVAULT_UV_IMAGE=ghcr.io/astral-sh/uv:0.4.28
+EVALVAULT_NODE_IMAGE=node:20.19-alpine
+EVALVAULT_NGINX_IMAGE=nginx:1.27.3-alpine
+
 # API_AUTH_TOKENS=
 # KNOWLEDGE_READ_TOKENS=
 # KNOWLEDGE_WRITE_TOKENS=

{evalvault-1.73.2 → evalvault-1.74.0}/Dockerfile

@@ -1,11 +1,16 @@
 # EvalVault Dockerfile
 # Multi-stage build for optimized production image
 
+# Stage 0: uv binary
+ARG PYTHON_IMAGE=python:3.12.6-slim
+ARG UV_IMAGE=ghcr.io/astral-sh/uv:0.4.28
+FROM ${UV_IMAGE} AS uv
+
 # Stage 1: Build stage
-FROM python:3.12-slim AS builder
+FROM ${PYTHON_IMAGE} AS builder
 
 # Install uv
-COPY --from=ghcr.io/astral-sh/uv:latest /uv /usr/local/bin/uv
+COPY --from=uv /uv /usr/local/bin/uv
 
 # Set working directory
 WORKDIR /app
@@ -25,7 +30,8 @@ RUN uv sync --frozen --no-dev
 
 
 # Stage 2: Runtime stage
-FROM python:3.12-slim AS runtime
+ARG PYTHON_IMAGE=python:3.12.6-slim
+FROM ${PYTHON_IMAGE} AS runtime
 
 # Create non-root user for security
 RUN useradd --create-home --shell /bin/bash evalvault

{evalvault-1.73.2 → evalvault-1.74.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: evalvault
-Version: 1.73.2
+Version: 1.74.0
 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

{evalvault-1.73.2 → evalvault-1.74.0}/docker-compose.offline.yml

@@ -17,8 +17,9 @@
 services:
   # PostgreSQL database for evaluation storage
   postgres:
-    image: postgres:16-alpine
+    image: ${POSTGRES_IMAGE:-postgres:16.4-alpine}
     container_name: evalvault-postgres
+    pull_policy: never
     restart: unless-stopped
     environment:
       POSTGRES_USER: ${POSTGRES_USER:-evalvault}
@@ -40,8 +41,12 @@ services:
     build:
       context: .
       dockerfile: Dockerfile
+      args:
+        PYTHON_IMAGE: ${EVALVAULT_PYTHON_IMAGE:-python:3.12.6-slim}
+        UV_IMAGE: ${EVALVAULT_UV_IMAGE:-ghcr.io/astral-sh/uv:0.4.28}
     container_name: evalvault-api
     restart: unless-stopped
+    pull_policy: never
     depends_on:
       postgres:
         condition: service_healthy
@@ -96,8 +101,12 @@ services:
     build:
       context: ./frontend
       dockerfile: Dockerfile
+      args:
+        NODE_IMAGE: ${EVALVAULT_NODE_IMAGE:-node:20.19-alpine}
+        NGINX_IMAGE: ${EVALVAULT_NGINX_IMAGE:-nginx:1.27.3-alpine}
     container_name: evalvault-web
     restart: unless-stopped
+    pull_policy: never
     depends_on:
       evalvault-api:
         condition: service_healthy

{evalvault-1.73.2 → evalvault-1.74.0}/docs/guides/OFFLINE_DOCKER.md

@@ -16,6 +16,7 @@ EvalVault를 외부망 없이 운영하기 위한 **오프라인 Docker 패키
 - `frontend/Dockerfile`: Web UI 정적 서빙 이미지
 - `frontend/nginx.conf`: `/api/*` 프록시 + SPA 라우팅
 - `scripts/offline/*.sh`: 이미지 export/import/smoke-test
+ - `.env.offline.example`: 오프라인 빌드용 베이스 이미지 고정
 
 ## 1) 환경 파일 준비
 
@@ -93,8 +94,22 @@ chmod +x scripts/offline/*.sh
 ./scripts/offline/export_images.sh
 ```
 
-- 산출물: `dist/evalvault_offline.tar`
-- 체크섬: `dist/evalvault_offline.tar.sha256`
+- 산출물: `dist/evalvault_offline_<timestamp>.tar`
+- 체크섬: `dist/evalvault_offline_<timestamp>.tar.sha256`
+
+파일명을 고정하려면 `OUTPUT_TAR`를 지정하세요.
+
+```bash
+OUTPUT_TAR=dist/evalvault_offline_legacy.tar ./scripts/offline/export_images.sh
+```
+
+이미지 태그를 고정하려면 `.env.offline` 또는 환경 변수로 다음을 지정합니다.
+
+- `EVALVAULT_PYTHON_IMAGE`
+- `EVALVAULT_UV_IMAGE`
+- `EVALVAULT_NODE_IMAGE`
+- `EVALVAULT_NGINX_IMAGE`
+- `POSTGRES_IMAGE` (옵션)
 
 Postgres 이미지를 함께 포함하려면:
 
@@ -114,6 +129,8 @@ INCLUDE_POSTGRES=1 ./scripts/offline/export_images.sh
 docker compose --env-file .env.offline -f docker-compose.offline.yml up -d
 ```
 
+주의: 폐쇄망에서는 외부 레지스트리 접근이 불가하므로, 반드시 `import_images.sh`로 이미지를 로드한 뒤 실행해야 합니다.
+
 - API: `http://localhost:8000`
 - Web UI: `http://localhost:8080`
 

{evalvault-1.73.2 → evalvault-1.74.0}/frontend/Dockerfile

@@ -1,6 +1,8 @@
 # syntax=docker/dockerfile:1
 
-FROM node:20-alpine AS build
+ARG NODE_IMAGE=node:20.19-alpine
+ARG NGINX_IMAGE=nginx:1.27.3-alpine
+FROM ${NODE_IMAGE} AS build
 
 WORKDIR /app
 
@@ -10,7 +12,7 @@ RUN npm ci
 COPY . ./
 RUN npm run build
 
-FROM nginx:1.27-alpine
+FROM ${NGINX_IMAGE}
 
 COPY nginx.conf /etc/nginx/conf.d/default.conf
 COPY --from=build /app/dist /usr/share/nginx/html

evalvault-1.74.0/frontend/src/components/ToastProvider.tsx

@@ -0,0 +1,89 @@
+import { createContext, useCallback, useContext, useEffect, useMemo, useRef, useState } from "react";
+
+type ToastVariant = "info" | "success" | "warning" | "error";
+
+type ToastItem = {
+    id: string;
+    message: string;
+    variant: ToastVariant;
+};
+
+type ToastOptions = {
+    variant?: ToastVariant;
+    durationMs?: number;
+};
+
+type ToastContextValue = {
+    pushToast: (message: string, options?: ToastOptions) => void;
+};
+
+const ToastContext = createContext<ToastContextValue | null>(null);
+
+export function ToastProvider({ children }: { children: React.ReactNode }) {
+    const [toasts, setToasts] = useState<ToastItem[]>([]);
+    const timersRef = useRef<Map<string, number>>(new Map());
+    const counterRef = useRef(0);
+
+    const removeToast = useCallback((id: string) => {
+        setToasts((prev) => prev.filter((toast) => toast.id !== id));
+        const timer = timersRef.current.get(id);
+        if (timer) {
+            window.clearTimeout(timer);
+            timersRef.current.delete(id);
+        }
+    }, []);
+
+    const pushToast = useCallback((message: string, options?: ToastOptions) => {
+        const id = `${Date.now()}-${counterRef.current++}`;
+        const variant = options?.variant ?? "info";
+        const durationMs = options?.durationMs ?? 2200;
+
+        setToasts((prev) => [...prev, { id, message, variant }]);
+        const timer = window.setTimeout(() => removeToast(id), durationMs);
+        timersRef.current.set(id, timer);
+    }, [removeToast]);
+
+    useEffect(() => () => {
+        timersRef.current.forEach((timer) => window.clearTimeout(timer));
+        timersRef.current.clear();
+    }, []);
+
+    const contextValue = useMemo(() => ({ pushToast }), [pushToast]);
+
+    return (
+        <ToastContext.Provider value={contextValue}>
+            {children}
+            <div className="fixed right-6 top-6 z-50 space-y-2">
+                {toasts.map((toast) => (
+                    <div
+                        key={toast.id}
+                        className={`rounded-lg border px-3 py-2 text-xs shadow-md backdrop-blur bg-background/95 ${variantClass(toast.variant)}`}
+                    >
+                        {toast.message}
+                    </div>
+                ))}
+            </div>
+        </ToastContext.Provider>
+    );
+}
+
+export function useToast() {
+    const context = useContext(ToastContext);
+    if (!context) {
+        throw new Error("useToast must be used within ToastProvider");
+    }
+    return context;
+}
+
+function variantClass(variant: ToastVariant) {
+    if (variant === "success") {
+        return "border-emerald-500/30 text-emerald-600 bg-emerald-500/10";
+    }
+    if (variant === "warning") {
+        return "border-amber-500/30 text-amber-600 bg-amber-500/10";
+    }
+    if (variant === "error") {
+        return "border-rose-500/30 text-rose-600 bg-rose-500/10";
+    }
+    return "border-border text-muted-foreground";
+}

evalvault-1.74.0/frontend/src/pages/AiSdkChat.tsx

@@ -0,0 +1,282 @@
+import { useEffect, useMemo, useState, type FormEvent } from "react";
+import { MessageSquare } from "lucide-react";
+import { useChat } from "@ai-sdk/react";
+import { DefaultChatTransport, type UIMessage } from "ai";
+import { Layout } from "../components/Layout";
+import { Conversation, Message, PromptInput } from "../components/ai-elements";
+import { fetchRuns, type RunSummary } from "../services/api";
+
+type ChatDataParts = {
+    status: { message: string };
+};
+
+type ChatUiMessage = UIMessage<unknown, ChatDataParts>;
+
+type ChatCategory = "dataset" | "analysis_method" | "result_interpretation" | "improvement_direction";
+
+const categoryOptions: { value: ChatCategory; label: string; hint: string }[] = [
+    {
+        value: "dataset",
+        label: "데이터셋 질문",
+        hint: "데이터셋 구성/필드/전처리 관련",
+    },
+    {
+        value: "analysis_method",
+        label: "분석 방식",
+        hint: "메트릭/평가/분석 흐름 질문",
+    },
+    {
+        value: "result_interpretation",
+        label: "결과 해석",
+        hint: "특정 run 결과 요약/해석",
+    },
+    {
+        value: "improvement_direction",
+        label: "개선 방향",
+        hint: "특정 run 개선/다음 액션",
+    },
+];
+
+const getMessageText = (message: ChatUiMessage) => {
+    return message.parts
+        .filter((part) => part.type === "text")
+        .map((part) => part.text)
+        .join("")
+        .trim();
+};
+
+export function AiSdkChat() {
+    const [input, setInput] = useState("");
+    const [statusMessage, setStatusMessage] = useState<string | null>(null);
+    const [runs, setRuns] = useState<RunSummary[]>([]);
+    const [runsError, setRunsError] = useState<string | null>(null);
+    const [runId, setRunId] = useState("");
+    const [category, setCategory] = useState<ChatCategory>("result_interpretation");
+
+    const trimmedRunId = runId.trim();
+    const selectedRun = useMemo(
+        () => runs.find((run) => run.run_id === trimmedRunId),
+        [runs, trimmedRunId]
+    );
+    const hasRun = Boolean(selectedRun);
+    const hasCases = (selectedRun?.total_test_cases ?? 0) > 0;
+    const hasMetrics = (selectedRun?.metrics_evaluated?.length ?? 0) > 0;
+    const isUnknownRun = Boolean(trimmedRunId) && !hasRun;
+
+    const categoryAvailability = useMemo(
+        () => ({
+            dataset: { enabled: true, reason: "" },
+            analysis_method: {
+                enabled: !trimmedRunId || hasMetrics,
+                reason: trimmedRunId && !hasMetrics ? "메트릭 정보 없음" : "",
+            },
+            result_interpretation: {
+                enabled: hasRun && hasCases,
+                reason: !trimmedRunId
+                    ? "run 선택 필요"
+                    : !hasRun
+                      ? "run_id 확인 필요"
+                      : "테스트 케이스 없음",
+            },
+            improvement_direction: {
+                enabled: hasRun && hasCases && hasMetrics,
+                reason: !trimmedRunId
+                    ? "run 선택 필요"
+                    : !hasRun
+                      ? "run_id 확인 필요"
+                      : !hasCases
+                        ? "테스트 케이스 없음"
+                        : "메트릭 정보 없음",
+            },
+        }),
+        [trimmedRunId, hasRun, hasCases, hasMetrics]
+    );
+
+    useEffect(() => {
+        if (categoryAvailability[category].enabled) {
+            return;
+        }
+        const next = categoryOptions.find((option) => categoryAvailability[option.value].enabled);
+        if (next) {
+            setCategory(next.value);
+        }
+    }, [category, categoryAvailability]);
+
+    useEffect(() => {
+        let active = true;
+        fetchRuns()
+            .then((items) => {
+                if (!active) return;
+                setRuns(items);
+            })
+            .catch((err) => {
+                if (!active) return;
+                setRunsError(err instanceof Error ? err.message : "run 목록을 불러오지 못했습니다.");
+            });
+        return () => {
+            active = false;
+        };
+    }, []);
+
+    const transport = useMemo(() => {
+        return new DefaultChatTransport({
+            api: "/api/v1/chat/ai-stream",
+            body: () => ({
+                run_id: trimmedRunId ? trimmedRunId : null,
+                category,
+            }),
+        });
+    }, [runId, category]);
+
+    const { messages, sendMessage, status, error, stop } = useChat<ChatUiMessage>({
+        transport,
+        onData: (dataPart) => {
+            if (dataPart.type === "data-status" && typeof dataPart.data.message === "string") {
+                setStatusMessage(dataPart.data.message);
+            }
+        },
+        onError: (err) => {
+            setStatusMessage(err.message);
+        },
+    });
+
+    const isLoading = status === "submitted" || status === "streaming";
+
+    const handleSubmit = (event: FormEvent<HTMLFormElement>) => {
+        event.preventDefault();
+        if (!input.trim() || isLoading) {
+            return;
+        }
+        if (!categoryAvailability[category].enabled) {
+            setStatusMessage("선택한 분류를 사용할 수 없습니다. 다른 분류를 선택해주세요.");
+            return;
+        }
+        const text = input.trim();
+        setInput("");
+        setStatusMessage(null);
+        sendMessage(
+            { text },
+            {
+                body: {
+                    run_id: runId.trim() ? runId.trim() : null,
+                    category,
+                },
+            }
+        );
+    };
+
+    const handleStop = () => {
+        stop();
+    };
+
+    return (
+        <Layout>
+            <div className="-m-4 lg:-m-8 flex h-[calc(100vh-6rem)] flex-col">
+                <div className="px-4 lg:px-8 pt-4 lg:pt-6">
+                    <div className="rounded-2xl border border-border/60 bg-background/70 backdrop-blur-sm p-4 lg:p-5">
+                        <div className="flex flex-col gap-3 lg:flex-row lg:items-center lg:justify-between">
+                            <div className="text-sm text-muted-foreground">
+                                <p className="font-semibold text-foreground">대상 run과 질문 분류를 먼저 선택하세요.</p>
+                                <p className="text-xs">선택 정보는 서버에 함께 전송됩니다.</p>
+                            </div>
+                            {runsError && (
+                                <div className="text-xs text-destructive">{runsError}</div>
+                            )}
+                        </div>
+                        <div className="mt-4 grid gap-4 lg:grid-cols-2">
+                            <div className="space-y-2">
+                                <label className="text-xs font-semibold text-muted-foreground">run_id 선택</label>
+                                <div className="flex flex-col gap-2">
+                                    <select
+                                        value={runId}
+                                        onChange={(event) => setRunId(event.target.value)}
+                                        className="w-full rounded-lg border border-border/60 bg-background/60 px-3 py-2 text-sm focus:outline-none focus:ring-1 focus:ring-primary/30"
+                                    >
+                                        <option value="">선택 안 함</option>
+                                        {runs.slice(0, 20).map((run) => (
+                                            <option key={run.run_id} value={run.run_id}>
+                                                {run.run_id} · {run.dataset_name} · {run.model_name}
+                                            </option>
+                                        ))}
+                                    </select>
+                                    <input
+                                        value={runId}
+                                        onChange={(event) => setRunId(event.target.value)}
+                                        placeholder="run_id를 직접 입력하세요"
+                                        className="w-full rounded-lg border border-border/60 bg-background/60 px-3 py-2 text-sm focus:outline-none focus:ring-1 focus:ring-primary/30"
+                                    />
+                                </div>
+                            </div>
+                            <div className="space-y-2">
+                                <label className="text-xs font-semibold text-muted-foreground">질문 분류</label>
+                                    <select
+                                        value={category}
+                                        onChange={(event) => setCategory(event.target.value as ChatCategory)}
+                                        className="w-full rounded-lg border border-border/60 bg-background/60 px-3 py-2 text-sm focus:outline-none focus:ring-1 focus:ring-primary/30"
+                                    >
+                                        {categoryOptions.map((option) => (
+                                            <option
+                                                key={option.value}
+                                                value={option.value}
+                                                disabled={!categoryAvailability[option.value].enabled}
+                                            >
+                                                {option.label}
+                                                {!categoryAvailability[option.value].enabled
+                                                    ? ` · ${categoryAvailability[option.value].reason}`
+                                                    : ""}
+                                            </option>
+                                        ))}
+                                    </select>
+                                    <p className="text-xs text-muted-foreground">
+                                        {categoryOptions.find((option) => option.value === category)?.hint}
+                                    </p>
+                                    {isUnknownRun && (
+                                        <p className="text-xs text-destructive">
+                                            선택한 run_id가 목록에 없습니다. run 의존 분류는 비활성화됩니다.
+                                        </p>
+                                    )}
+                            </div>
+                        </div>
+                    </div>
+                </div>
+                <div className="flex-1 relative bg-background/50">
+                    {messages.length === 0 ? (
+                        <div className="h-full flex flex-col items-center justify-center text-center p-8 space-y-4 opacity-70">
+                            <div className="w-16 h-16 rounded-2xl bg-primary/10 flex items-center justify-center mb-4">
+                                <MessageSquare className="w-8 h-8 text-primary" />
+                            </div>
+                            <h3 className="text-lg font-semibold text-foreground">AI SDK 채팅을 시작해보세요</h3>
+                            <p className="text-sm text-muted-foreground max-w-sm">
+                                AI SDK 스트리밍 프로토콜로 동작하는 챗봇입니다. 분석 결과 해석과 개선 아이디어를
+                                바로 질문해볼 수 있습니다.
+                            </p>
+                        </div>
+                    ) : (
+                        <Conversation>
+                            {messages.map((msg) => {
+                                const content = getMessageText(msg);
+                                if (!content) {
+                                    return null;
+                                }
+                                const role = msg.role === "user" ? "user" : "assistant";
+                                return <Message key={msg.id} role={role} content={content} />;
+                            })}
+                            {(statusMessage || error) && (
+                                <div className="p-4 rounded-lg bg-secondary/60 border border-border/60 text-sm text-muted-foreground text-center">
+                                    {statusMessage ?? error?.message}
+                                </div>
+                            )}
+                        </Conversation>
+                    )}
+                </div>
+                <PromptInput
+                    input={input}
+                    handleInputChange={(event) => setInput(event.target.value)}
+                    handleSubmit={handleSubmit}
+                    isLoading={isLoading}
+                    stop={handleStop}
+                />
+            </div>
+        </Layout>
+    );
+}

{evalvault-1.73.2 → evalvault-1.74.0}/frontend/src/services/api.ts

@@ -352,6 +352,25 @@ export interface DebugReport {
     langfuse_trace_url?: string | null;
 }
 
+export interface OpsReportResponse {
+    run_summary: Record<string, unknown>;
+    ops_kpis: {
+        total_test_cases?: number | null;
+        pass_rate?: number | null;
+        failure_rate?: number | null;
+        stage_error_rate?: number | null;
+        stage_error_severity?: "ok" | "warning" | "critical" | string | null;
+        duration_seconds?: number | null;
+        total_tokens?: number | null;
+        total_cost_usd?: number | null;
+        avg_latency_ms?: number | null;
+        p95_latency_ms?: number | null;
+        avg_tokens_per_case?: number | null;
+        avg_cost_per_case_usd?: number | null;
+    };
+    metadata?: Record<string, unknown>;
+}
+
 export interface PromptDiffSummaryItem {
     role: string;
     base_checksum?: string | null;
@@ -832,6 +851,14 @@ export async function fetchDebugReportMarkdown(runId: string): Promise<Blob> {
     return response.blob();
 }
 
+export async function fetchOpsReport(runId: string): Promise<OpsReportResponse> {
+    const response = await fetch(`${API_BASE_URL}/runs/${runId}/ops-report?format=json&save=false`);
+    if (!response.ok) {
+        throw new Error(`Failed to fetch ops report: ${response.statusText}`);
+    }
+    return response.json();
+}
+
 export async function fetchPromptDiff(
     baseRunId: string,
     targetRunId: string,

{evalvault-1.73.2 → evalvault-1.74.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "evalvault"
-version = "1.73.2"
+version = "1.74.0"
 description = "RAG evaluation system using Ragas with Phoenix/Langfuse tracing"
 readme = "README.md"
 requires-python = ">=3.12"

evalvault-1.74.0/scripts/offline/export_base_images.sh

@@ -0,0 +1,54 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+# 베이스 이미지 export 스크립트
+# 폐쇄망에서 빌드하기 위해 필요한 모든 베이스 이미지를 export합니다.
+
+ROOT_DIR=$(cd "$(dirname "${BASH_SOURCE[0]}")/../.." && pwd)
+cd "$ROOT_DIR"
+
+OUTPUT_TAR=${OUTPUT_TAR:-dist/evalvault_base_images.tar}
+INCLUDE_POSTGRES=${INCLUDE_POSTGRES:-1}
+
+# .env.offline에서 이미지 태그 읽기 (없으면 기본값 사용)
+if [ -f .env.offline ]; then
+  source .env.offline
+fi
+
+BASE_IMAGES=(
+  "${EVALVAULT_PYTHON_IMAGE:-python:3.12.6-slim}"
+  "${EVALVAULT_UV_IMAGE:-ghcr.io/astral-sh/uv:0.4.28}"
+  "${EVALVAULT_NODE_IMAGE:-node:20.11-alpine}"
+  "${EVALVAULT_NGINX_IMAGE:-nginx:1.27.3-alpine}"
+)
+
+if [ "$INCLUDE_POSTGRES" = "1" ]; then
+  BASE_IMAGES+=("${POSTGRES_IMAGE:-postgres:16.4-alpine}")
+fi
+
+echo "다음 베이스 이미지들을 pull합니다:"
+for img in "${BASE_IMAGES[@]}"; do
+  echo "  - $img"
+done
+
+# 모든 베이스 이미지 pull
+for img in "${BASE_IMAGES[@]}"; do
+  echo "Pulling $img..."
+  docker pull "$img"
+done
+
+# 이미지들을 하나의 tar 파일로 저장
+mkdir -p "$(dirname "$OUTPUT_TAR")"
+echo "베이스 이미지들을 저장합니다: $OUTPUT_TAR"
+docker save -o "$OUTPUT_TAR" "${BASE_IMAGES[@]}"
+
+# 체크섬 생성
+sha256sum "$OUTPUT_TAR" > "${OUTPUT_TAR}.sha256"
+
+echo ""
+echo "✅ 완료!"
+echo "저장된 파일: $OUTPUT_TAR"
+echo "체크섬: ${OUTPUT_TAR}.sha256"
+echo ""
+echo "폐쇄망에서 로드하려면:"
+echo "  docker load -i $OUTPUT_TAR"