oh-my-knowledge 0.31.0 → 0.33.0

package/dist/renderer/layout.js

@@ -0,0 +1,776 @@
+export function e(text) {
+    return String(text)
+        .replaceAll('&', '&')
+        .replaceAll('<', '&lt;')
+        .replaceAll('>', '&gt;')
+        .replaceAll('"', '&quot;')
+        .replaceAll("'", '&#39;');
+}
+export function fmtNum(n, digits = 0) {
+    return Number(n || 0).toLocaleString('en-US', { maximumFractionDigits: digits });
+}
+export function fmtDuration(ms) {
+    const v = Number(ms || 0);
+    if (v < 1000)
+        return `${v}ms`;
+    if (v < 60000)
+        return `${(v / 1000).toFixed(1)}s`;
+    const min = Math.floor(v / 60000);
+    const sec = Math.round((v % 60000) / 1000);
+    return sec > 0 ? `${min}m${sec}s` : `${min}m`;
+}
+export function fmtCost(usd, reported = true) {
+    // reported=false 时 executor 不报 cost(如 codex CLI),`usd` 是占位 0,
+    // 显示 "—" 跟"真的花了 $0"区分开。callsite 传 reported 时通常来自
+    // `VariantSummary.execCostReported !== false` 或 `VariantResult.costReportedByExecutor !== false`。
+    if (!reported)
+        return '—';
+    return `$${Number(usd || 0).toFixed(4)}`;
+}
+export function fmtKnownCost(usd, fullyReported = true) {
+    const value = Number(usd || 0);
+    if (fullyReported)
+        return fmtCost(value);
+    return value > 0 ? `≥${fmtCost(value)}` : '—';
+}
+export function fmtLocalTime(isoStr) {
+    const d = new Date(isoStr);
+    const pad = (n) => String(n).padStart(2, '0');
+    return `${d.getFullYear()}-${pad(d.getMonth() + 1)}-${pad(d.getDate())} ${pad(d.getHours())}:${pad(d.getMinutes())}:${pad(d.getSeconds())}`;
+}
+export function delta(a, b, lowerIsBetter = false) {
+    if (!a || !b || a === 0)
+        return '';
+    const pct = ((b - a) / a * 100).toFixed(1);
+    const better = lowerIsBetter ? b < a : b > a;
+    const color = better ? 'var(--green)' : b === a ? 'var(--text-muted)' : 'var(--red)';
+    const arrow = b > a ? '↑' : b < a ? '↓' : '→';
+    return `<span style="color:${color};font-size:11px;margin-left:4px">${arrow}${Math.abs(Number(pct))}%</span>`;
+}
+export const COLORS = ['var(--chart-1)', 'var(--chart-2)', 'var(--chart-3)', 'var(--chart-4)', 'var(--chart-5)', 'var(--chart-6)'];
+export const I18N = {
+    zh: {
+        title: '评测报告',
+        subtitle: '知识载体评测',
+        noRuns: '暂无评测记录。运行 <code>omk eval --control v1 --treatment v2</code> 开始。',
+        runId: '报告名称', variants: '实验分组', model: '任务执行模型', samples: '评测用例数',
+        score: '分数', cost: '执行成本', time: '时间',
+        deleteBtnText: '删除', deleteConfirm: '确定删除报告', deleteFail: '删除失败',
+        reportTitle: '评测报告', backToList: '← 返回列表',
+        judge: '评委', executor: '执行器', blindLabel: '盲测', revealBlind: '显示变体对应关系',
+        dimFact: '📋 事实', dimFactDesc: '输出的事实声明是否正确（规则可验证：关键词匹配、格式校验等断言）',
+        dimBehavior: '🛠️ 行为', dimBehaviorDesc: '执行过程是否合规（规则可验证：工具调用路径、轮次限制、成本约束等断言）',
+        dimJudge: '💬 LLM 评价', dimJudgeDesc: '请一个 LLM 当评委，读任务执行模型的输出内容，按预先写好的评分规则（英文 rubric）打 1-5 分。主观但能抓到规则断言判不了的"整体好不好"',
+        dimQuality: '📊 质量', dimQualityDesc: '事实 + 行为 + LLM 评价的等权平均（1-5 分）。UI 已拆出三层平铺展示，composite 字段仅保留在 JSON 数据层',
+        dimCost: '💰 执行成本', dimCostDesc: '基于 Token 消耗量和模型定价计算的任务执行模型调用费用',
+        dimEfficiency: '⚡ 效率', dimEfficiencyDesc: 'Skill 从发送请求到模型返回完整响应的端到端耗时',
+        dimStability: '🛡️ 稳定性', dimStabilityDesc: '多次运行（--repeat ≥ 2）分数一致性的 CV 变异系数，单轮显示"—"',
+        compositeScore: '综合分数', scoreRange: '分数范围',
+        assertions: '断言', assertionsDesc: '规则检查得分：通过的断言权重占比映射到 1-5 分',
+        llmJudge: 'LLM 评委', llmJudgeDesc: '由评委模型按预先写好的评分规则（英文叫 rubric）打出的 1-5 分',
+        judgeStddev: '评委波动', judgeStddevDesc: '同一份输出让评委评 N 次 (--judge-repeat) 得到 N 个分数的标准差。值低 = 评委对自己很坚定；值高 = 这个分本身就是噪声',
+        judgeFailures: '评委失败', judgeFailuresDesc: 'N 次评委评分中返回 score=0（解析失败 / 调用错误）的次数。stddev=0 + failureCount>0 不是"完美一致"，是"大部分炸了"',
+        judgeReasoning: '评委推理', judgeReasoningExpand: '展开',
+        ensembleHeader: '多评委评分对比', ensembleDesc: '不同评委模型对同一份输出的独立评分。用于反驳"同模态偏差"',
+        agreementHeader: '跨用例评委一致性', agreementDesc: '在所有评测用例上算的多评委一致性',
+        pearsonLabel: '皮尔逊系数 (Pearson)', pearsonDesc: '皮尔逊相关系数：1=完全同向排序，0=无关，-1=完全反向',
+        madLabel: '平均绝对差 (MAD)', madDesc: '平均绝对差。1-5 制下 < 0.5 紧密一致, > 1.5 大分歧',
+        judgeModelsLabel: '评委模型',
+        judgeRepeatLabel: '每条用例评委评价次数',
+        judgePromptHashLabel: '评委提示词指纹', judgePromptHashDesc: '评委提示词模板的 SHA256 前 12 位。两份报告 hash 相同才能严格比分数',
+        sampleHashCount: '评测用例指纹', sampleHashCountDesc: '已记录内容指纹的评测用例数 / 全部评测用例数。每条评测用例算 SHA256 前 12 位，用于跨 run 识别"测的是不是同一件事"。两份报告对同一 sample_id hash 一致才能严格比分',
+        evalFrameworkLabel: '统计框架', evalFrameworkBootstrap: 'bootstrap CI', evalFrameworkBoth: 't-test + bootstrap', evalFrameworkTTest: 't-test',
+        evalFrameworkDesc: '分数置信区间用什么算法。bootstrap 不假设分布,适合 LLM 1-5 序数评分;t-test 假设正态,小样本下不稳。"both" = 报告同时含两种,renderer 优先 bootstrap',
+        bootstrapDiffSignificant: '✓ 显著差异', bootstrapDiffNotSignificant: '✗ 无显著差异', bootstrapDiffLabel: 'Δ (treatment - control)',
+        totalCost: '已上报总成本', inputTokens: '输入', outputTokens: '输出',
+        totalTokens: '总计', tokPerReq: 'tokens/次', avgLatency: '平均延迟',
+        successRate: '完成率', success: '成功', errors: '失败',
+        tokenComparison: 'Tokens 对比', latencyComparison: '延迟对比',
+        avgTurns: '平均轮次', turnsPerReq: '轮/次', minScore: '最低',
+        autoAnalysis: '自动分析',
+        perSampleDetail: '逐用例详情', sample: '用例',
+        scoreCol: '分数', tokensCol: 'Tokens', msCol: '延迟(ms)',
+        batchOverview: '总览', batchSkill: 'Skill', batchBaseline: '无 Skill', batchWithSkill: '有 Skill', batchImprovement: '提升',
+        batchSkills: '个 Skill', batchSamples: '个评测用例',
+        agentLabel: 'Agent 评测',
+        skillLabel: 'Skill 评测',
+        promptLabel: 'Prompt 评测',
+        workflowLabel: 'Workflow 评测',
+        agentOverview: 'Agent 执行概览',
+        agentToolCalls: '工具调用',
+        agentToolSuccess: '工具成功率',
+        agentToolDist: '工具分布',
+        agentAvgTurns: '平均轮次',
+        agentAvgTools: '平均工具调用',
+        traceToggle: '执行轨迹',
+        traceAssistant: '助手',
+        traceTool: '工具',
+        traceFullOutput: '查看完整输出',
+        traceExecMs: '执行',
+        traceGradeMs: '评分',
+        traceTotalMs: '总计',
+        variantConfig: '实验配置',
+        variantConfigDesc: '先看清楚在比较什么，再看分数。本表展示每个 variant 背后的 artifact 和运行环境。',
+        variantType: '实验类型',
+        variantArtifactKind: '知识类型',
+        variantArtifactSource: '知识来源',
+        variantExecutionStrategy: '执行策略',
+        variantRuntimeContext: '运行环境',
+        // --- observability (skill health / trend / diff) ---
+        skillHealthTitle: 'Skill 健康度日报',
+        noAnalyses: '暂无 skill 健康度日报。运行 <code>omk observe &lt;trace-dir&gt;</code> 生成。',
+        backToEvalReports: '← 评测报告',
+        backToAnalyses: '← Skill 健康度日报',
+        analysesCompareHint: '选两个报告的 from/to 单选框,点 Compare 生成 diff。',
+        analysesCompareBtn: '对比 →',
+        analysesFromLabel: 'from',
+        analysesToLabel: 'to',
+        analysesSessions: '会话',
+        analysesSegs: '段',
+        analysesSkills: '技能',
+        analysesLowN: '样本不足',
+        skillTrendHeading: 'Skill 趋势',
+        noTrendData: '暂无趋势数据。该 skill 尚未出现在任何分析报告里。',
+        trendNPoints: '个时间点',
+        trendEarliest: '最早',
+        trendLatest: '最新',
+        trendLegendGap: 'gap rate',
+        trendLegendWeighted: 'weighted gap',
+        trendLegendFailure: 'failure rate',
+        trendLegendCoverage: 'coverage',
+        trendColTimestamp: '时间',
+        trendColSegs: '段数',
+        trendColGap: 'Gap',
+        trendColWeighted: '加权',
+        trendColFailure: '失败率',
+        trendColCoverage: '覆盖',
+        trendColTokens: 'Tokens',
+        trendColDuration: '耗时',
+        skillDiffHeading: 'Skill 健康度对比',
+        diffSortHint: '按 gap 变化量排序;绿色=改善,红色=恶化',
+        diffTagRemoved: '已消失',
+        diffTagNew: '新增',
+        diffNavFrom: '起点',
+        diffNavTo: '终点',
+        diffColSkill: 'Skill',
+        diffColSegments: '段数',
+        diffColWeightedGap: '加权 Gap',
+        diffColFailureRate: '失败率',
+        diffColCoverage: '覆盖',
+        viewTrendLink: '查看趋势 →',
+        artifactHashLabel: '版本指纹',
+        artifactHashTooltip: 'skill 文件内容的 SHA-256 前 12 位(不含路径/时间/git),用于辨别报告对应哪一版 skill;同文件多次跑指纹不变,改一字节就变——防止"改动效果"和"随机波动"混淆',
+        switchLang: '英文',
+    },
+    en: {
+        title: 'Evaluation Reports',
+        subtitle: 'Knowledge Artifact Evaluation',
+        noRuns: 'No evaluation runs yet. Run <code>omk eval --control v1 --treatment v2</code> to start.',
+        runId: 'Report', variants: 'Variant', model: 'Task execution model', samples: 'Samples',
+        score: 'Score', cost: 'Execution cost', time: 'Time',
+        deleteBtnText: 'Delete', deleteConfirm: 'Delete report', deleteFail: 'Delete failed',
+        reportTitle: 'Evaluation Report', backToList: '← Back to list',
+        judge: 'Judge', executor: 'Executor', blindLabel: 'BLIND', revealBlind: 'Reveal variant mapping',
+        dimFact: '📋 Fact', dimFactDesc: 'Are factual claims correct (rule-verified: keyword matching, schema checks, etc.)',
+        dimBehavior: '🛠️ Behavior', dimBehaviorDesc: 'Is execution compliant (rule-verified: tool paths, turn limits, cost constraints)',
+        dimJudge: '💬 LLM judge', dimJudgeDesc: 'A separate LLM acts as judge: reads the task execution model output, scores 1-5 against a predefined rubric. Subjective, catches "overall feel" rule-based assertions miss',
+        dimQuality: '📊 Quality', dimQualityDesc: 'Equal-weight average of Fact + Behavior + LLM judge (1-5). UI now splits the three layers into separate columns; composite lives only in JSON data',
+        dimCost: '💰 Exec cost', dimCostDesc: 'API cost for the task execution model, calculated from token usage and model pricing',
+        dimEfficiency: '⚡ Efficiency', dimEfficiencyDesc: 'End-to-end latency from sending request to receiving full response',
+        dimStability: '🛡️ Stability', dimStabilityDesc: 'How much the score swings across repeated runs. Needs `--repeat ≥ 2`; single-run shows "—" because stability cannot be measured from one run',
+        compositeScore: 'composite score', scoreRange: 'Range',
+        assertions: 'Assertions', assertionsDesc: 'Rule-based score: passed assertion weight ratio mapped to 1-5',
+        llmJudge: 'LLM Judge', llmJudgeDesc: 'Score (1-5) from the judge model based on a predefined rubric (scoring criteria)',
+        judgeStddev: 'Judge stddev', judgeStddevDesc: 'Stddev across N judge calls (--judge-repeat). Low = judge is consistent; high = this score itself is noisy',
+        judgeFailures: 'Judge failures', judgeFailuresDesc: 'How many of N judge calls returned score=0 (parse / executor failure). stddev=0 + failureCount>0 is NOT "perfect agreement" — it means most calls failed',
+        judgeReasoning: 'CoT reasoning', judgeReasoningExpand: 'expand',
+        ensembleHeader: 'Per-judge scores', ensembleDesc: 'Independent scores from different judge models for the same output — refutes same-modality bias',
+        agreementHeader: 'Inter-judge agreement', agreementDesc: 'Cross-sample agreement metrics across all judges in this variant',
+        pearsonLabel: 'Pearson', pearsonDesc: 'Pearson correlation: 1=perfect rank agreement, 0=independent, -1=anti-correlated',
+        madLabel: 'MAD', madDesc: 'Mean absolute difference. On 1-5 scale: < 0.5 tight, > 1.5 large disagreement',
+        judgeModelsLabel: 'Judge models',
+        judgeRepeatLabel: 'Judge calls per sample',
+        judgePromptHashLabel: 'Judge prompt fingerprint', judgePromptHashDesc: 'SHA256-12 of the judge prompt template. Reports must share the same hash before scores are strictly comparable',
+        sampleHashCount: 'Sample fingerprints', sampleHashCountDesc: 'Number of samples with content hashes recorded / total samples. Each sample gets a SHA256-12 fingerprint identifying "is this the same sample as before". Two reports must share the same hash for a given sample_id to be strictly comparable',
+        evalFrameworkLabel: 'CI framework', evalFrameworkBootstrap: 'bootstrap CI', evalFrameworkBoth: 't-test + bootstrap', evalFrameworkTTest: 't-test',
+        evalFrameworkDesc: 'Algorithm used for confidence intervals. Bootstrap is distribution-free and recommended for ordinal LLM scores; t-test assumes normality and is unstable on small N. "both" = report has both, renderer prefers bootstrap',
+        bootstrapDiffSignificant: '✓ significant', bootstrapDiffNotSignificant: '✗ not significant', bootstrapDiffLabel: 'Δ (treatment - control)',
+        totalCost: 'Reported total cost', inputTokens: 'Input', outputTokens: 'Output',
+        totalTokens: 'Total', tokPerReq: 'tokens/req', avgLatency: 'avg latency',
+        successRate: 'completion rate', success: 'Success', errors: 'Errors',
+        tokenComparison: 'Tokens Comparison', latencyComparison: 'Latency Comparison',
+        avgTurns: 'Avg Turns', turnsPerReq: 'turns/req', minScore: 'Min',
+        autoAnalysis: 'Auto Analysis',
+        perSampleDetail: 'Per-Sample Detail', sample: 'Sample',
+        scoreCol: 'Score', tokensCol: 'Tokens', msCol: 'ms',
+        batchOverview: 'Overview', batchSkill: 'Skill', batchBaseline: 'Baseline', batchWithSkill: 'With Skill', batchImprovement: 'Improvement',
+        batchSkills: ' skills', batchSamples: ' samples',
+        agentLabel: 'Agent Eval',
+        skillLabel: 'Skill Eval',
+        promptLabel: 'Prompt Eval',
+        workflowLabel: 'Workflow Eval',
+        agentOverview: 'Agent Execution Overview',
+        agentToolCalls: 'Tool Calls',
+        agentToolSuccess: 'Tool Success Rate',
+        agentToolDist: 'Tool Distribution',
+        agentAvgTurns: 'Avg Turns',
+        agentAvgTools: 'Avg Tool Calls',
+        traceToggle: 'Execution Trace',
+        traceAssistant: 'Assistant',
+        traceTool: 'Tool',
+        traceFullOutput: 'Full Output',
+        traceExecMs: 'Exec',
+        traceGradeMs: 'Grade',
+        traceTotalMs: 'Total',
+        variantConfig: 'Experiment Setup',
+        variantConfigDesc: 'Confirm what is being compared before reading the score. This table explains the artifact and runtime context behind each variant.',
+        variantType: 'Role',
+        variantArtifactKind: 'Artifact Kind',
+        variantArtifactSource: 'Source',
+        variantExecutionStrategy: 'Execution Strategy',
+        variantRuntimeContext: 'Runtime Context',
+        // --- observability (skill health / trend / diff) ---
+        skillHealthTitle: 'Skill Health Reports',
+        noAnalyses: 'No skill health reports yet. Run <code>omk observe &lt;trace-dir&gt;</code> to generate.',
+        backToEvalReports: '← Eval reports',
+        backToAnalyses: '← Skill Health Reports',
+        analysesCompareHint: 'Pick from/to radios on two reports, then click Compare to generate a diff.',
+        analysesCompareBtn: 'Compare →',
+        analysesFromLabel: 'from',
+        analysesToLabel: 'to',
+        analysesSessions: 'sessions',
+        analysesSegs: 'segs',
+        analysesSkills: 'skills',
+        analysesLowN: 'low N',
+        skillTrendHeading: 'Skill Trend',
+        noTrendData: 'No trend data. This skill has not appeared in any analysis report yet.',
+        trendNPoints: 'data points',
+        trendEarliest: 'earliest',
+        trendLatest: 'latest',
+        trendLegendGap: 'gap rate',
+        trendLegendWeighted: 'weighted gap',
+        trendLegendFailure: 'failure rate',
+        trendLegendCoverage: 'coverage',
+        trendColTimestamp: 'Timestamp',
+        trendColSegs: 'Segs',
+        trendColGap: 'Gap',
+        trendColWeighted: 'Weighted',
+        trendColFailure: 'Failure',
+        trendColCoverage: 'Coverage',
+        trendColTokens: 'Tokens',
+        trendColDuration: 'Duration',
+        skillDiffHeading: 'Skill Health Diff',
+        diffSortHint: 'Sorted by |Δgap|; green=improved, red=regressed',
+        diffTagRemoved: 'removed',
+        diffTagNew: 'new',
+        diffNavFrom: 'from',
+        diffNavTo: 'to',
+        diffColSkill: 'Skill',
+        diffColSegments: 'Segments',
+        diffColWeightedGap: 'Weighted gap',
+        diffColFailureRate: 'Failure rate',
+        diffColCoverage: 'Coverage',
+        viewTrendLink: 'trend →',
+        artifactHashLabel: 'Version fingerprint',
+        artifactHashTooltip: 'First 12 hex chars of SHA-256 over the skill file content (content-only: no path/time/git); identifies which version of the skill this report ran — same file = same fingerprint, any byte change = different fingerprint. Keeps "intentional change" separate from "random variance"',
+        switchLang: 'Chinese',
+    },
+};
+export const DEFAULT_LANG = 'zh';
+export function t(key, lang = DEFAULT_LANG) {
+    return I18N[lang]?.[key] || I18N.en[key] || key;
+}
+function globalKeyboardScript() {
+    return `
+  <script>
+  // Global modal helpers. openModal/closeModal manage focus so the close
+  // button becomes the first tab stop when a modal opens.
+  window.openModal = function(id) {
+    var modal = document.getElementById(id);
+    if (!modal) return;
+    modal.style.display = 'flex';
+    setTimeout(function() {
+      var closeBtn = modal.querySelector('.modal-close');
+      if (closeBtn) closeBtn.focus();
+    }, 30);
+  };
+  window.closeModal = function(id) {
+    var modal = document.getElementById(id);
+    if (!modal) return;
+    modal.style.display = 'none';
+  };
+  // Global ESC to close any open modal-overlay
+  document.addEventListener('keydown', function(e) {
+    if (e.key === 'Escape') {
+      document.querySelectorAll('.modal-overlay').forEach(function(m) {
+        if (m.style.display === 'flex') m.style.display = 'none';
+      });
+    }
+  });
+  // 复制 paste-ready 片段到剪贴板:供 insight modal 的"复制片段"按钮调用。
+  // 元素 id 指 <code> 节点;复制后按钮加 .copied 类做 1.5s 视觉反馈。
+  // i18n:按钮反馈文案随页面 data-lang(layout 写在 <html> 上,langToggleScript 切换时更新)。
+  // 未设 dataset 时按 DEFAULT_LANG(zh)兜底,跟旧报告/无 toggle 入口的页面一致。
+  window.omkCopySnippet = function(elementId, btn) {
+    var el = document.getElementById(elementId);
+    if (!el) return;
+    var text = el.textContent || '';
+    var pageLang = document.documentElement.dataset.lang || '${DEFAULT_LANG}';
+    var copiedLabel = pageLang === 'en' ? '✓ Copied' : '✓ 已复制';
+    var done = function() {
+      if (!btn) return;
+      var orig = btn.textContent;
+      btn.textContent = copiedLabel;
+      btn.classList.add('copied');
+      setTimeout(function() { btn.textContent = orig; btn.classList.remove('copied'); }, 1500);
+    };
+    if (navigator.clipboard && navigator.clipboard.writeText) {
+      navigator.clipboard.writeText(text).then(done, function() { /* swallow */ });
+    } else {
+      // 老浏览器兜底:execCommand fallback
+      var ta = document.createElement('textarea');
+      ta.value = text; ta.style.position = 'fixed'; ta.style.opacity = '0';
+      document.body.appendChild(ta); ta.select();
+      try { document.execCommand('copy'); done(); } catch(e) { /* swallow */ }
+      document.body.removeChild(ta);
+    }
+  };
+  </script>`;
+}
+function langToggleScript() {
+    return `
+  <script>
+  var I18N = ${JSON.stringify(I18N)};
+  function switchLang() {
+    var cur = document.documentElement.dataset.lang || '${DEFAULT_LANG}';
+    var next = cur === 'zh' ? 'en' : 'zh';
+    document.documentElement.dataset.lang = next;
+    document.documentElement.lang = next === 'zh' ? 'zh-CN' : 'en';
+    document.querySelectorAll('[data-i18n]').forEach(function(el) {
+      var key = el.dataset.i18n;
+      if (I18N[next][key]) {
+        if (el.tagName === 'INPUT') { el.placeholder = I18N[next][key]; }
+        else { el.innerHTML = I18N[next][key]; }
+      }
+    });
+    document.getElementById('lang-toggle').textContent = I18N[next].switchLang;
+    // 同步写入 URL ?lang= 和 localStorage,让刷新/跳转保持语言选择
+    try {
+      var url = new URL(window.location.href);
+      url.searchParams.set('lang', next);
+      window.history.replaceState(null, '', url.toString());
+      localStorage.setItem('omk-lang', next);
+    } catch (e) { /* ignore */ }
+  }
+  // 页面加载时,若 URL 无 lang 但 localStorage 有,跳转到带 lang 的 URL (仅一次)
+  (function() {
+    try {
+      var url = new URL(window.location.href);
+      if (!url.searchParams.get('lang')) {
+        var saved = localStorage.getItem('omk-lang');
+        if (saved && saved !== '${DEFAULT_LANG}') {
+          url.searchParams.set('lang', saved);
+          window.location.replace(url.toString());
+        }
+      }
+    } catch (e) { /* ignore */ }
+  })();
+  </script>`;
+}
+function langToggleButton(lang) {
+    return `<button id="lang-toggle" onclick="switchLang()" class="lang-toggle">${t('switchLang', lang)}</button>`;
+}
+export function layout(title, body, lang = DEFAULT_LANG) {
+    const htmlLang = lang === 'zh' ? 'zh-CN' : 'en';
+    const favicon = encodeURIComponent('<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 32 32"><defs><linearGradient id="g" x1="0" y1="0" x2="1" y2="1"><stop offset="0%" stop-color="#60a5fa"/><stop offset="100%" stop-color="#a78bfa"/></linearGradient></defs><circle cx="16" cy="16" r="15" fill="#0f172a"/><circle cx="16" cy="16" r="8" stroke="url(#g)" stroke-width="3.5" fill="none"/></svg>');
+    // 中英文切换按钮临时隐藏(URL ?lang= / localStorage 切换逻辑保留,按钮 UI 不渲染)。
+    // 想恢复:在 body 模板里加回 ${langToggleButton(lang)}。
+    void langToggleButton;
+    return `<!doctype html><html lang="${htmlLang}" data-lang="${lang}"><head><meta charset="utf-8"><meta name="viewport" content="width=device-width,initial-scale=1"><title>OMK · ${title}</title>
+<link rel="icon" type="image/svg+xml" href="data:image/svg+xml,${favicon}">${globalKeyboardScript()}
+<style>
+/* Pantone-inspired light theme(2026-05 全站浅色化)。
+   设计原则:暖白底 + 哑色调,不刺眼;字号 14-15px 起;subtle box-shadow 替代 border。
+   旧的深色蓝灰主题已迁移,语义保留但色值翻转。 */
+:root{
+  --bg-base:#fbfaf7;            /* 暖白纸感 */
+  --bg-surface:#ffffff;          /* 卡片白 */
+  --bg-elevated:rgba(58,58,58,.025);
+  --bg-soft:rgba(58,58,58,.04);
+  --border:#e8e3da;              /* 暖灰分隔 */
+  --border-hover:#d4ccbf;
+  --text-primary:#3a3a3a;        /* 深炭灰 */
+  --text-secondary:#7a7a7a;
+  --text-muted:#a8a8a8;
+  --text-faint:#c8c8c8;
+  --accent:#5a7a93;              /* 板岩蓝(slate) */
+  --accent-hover:#3f5d76;
+  --green:#5e8252;               /* 鼠尾草绿(sage) */
+  --green-bg:rgba(94,130,82,.10);
+  --red:#9c4a3f;                 /* 砖红土陶(terracotta) */
+  --red-bg:rgba(156,74,63,.10);
+  --yellow:#b08030;              /* 暖琥珀 */
+  --yellow-bg:rgba(176,128,48,.10);
+  --info-bg:rgba(90,122,147,.08);
+  /* 图表色 — 全部哑色调,跨色相区分但饱和度统一在 ~50% */
+  --chart-1:#5a7a93;
+  --chart-2:#7a6b89;
+  --chart-3:#5e8252;
+  --chart-4:#b08030;
+  --chart-5:#9c6478;
+  --chart-6:#937846;
+  --bg-card:#ffffff;
+  --radius:8px;
+  --radius-lg:12px;
+  /* 字号上调:正文 14px / 标签 13px / 详情 13px / 微元数据 12px */
+  --fs-micro:12px;
+  --fs-detail:13px;
+  --fs-label:13px;
+  --fs-body:14.5px;
+  --shadow-sm:0 1px 3px rgba(58,58,58,.06);
+  --shadow-md:0 2px 8px rgba(58,58,58,.10);
+}
+*{box-sizing:border-box;margin:0}
+body{font-family:-apple-system,BlinkMacSystemFont,"PingFang SC","Microsoft YaHei","Segoe UI",Roboto,sans-serif;padding:32px;background:var(--bg-base);color:var(--text-primary);min-height:100vh;line-height:1.7;max-width:1440px;margin:0 auto;font-size:var(--fs-body)}
+h1{margin:0 0 8px;font-size:1.75rem;font-weight:600;color:var(--text-primary);letter-spacing:-0.01em;line-height:1.3}
+h2{margin:32px 0 12px;font-size:1.0625rem;color:var(--text-primary);font-weight:600;line-height:1.4}
+.subtitle{color:var(--text-secondary);font-size:0.875rem;margin:0 0 24px}
+a{color:var(--accent);text-decoration:none;transition:color .15s}
+a:hover{color:var(--accent-hover);text-decoration:underline}
+
+/* Meta tags — 浅色主题下加大对比度,小一点的字号也清晰可读 */
+.meta-tags{display:flex;flex-wrap:wrap;gap:8px;margin:8px 0 20px}
+.meta-tag{font-size:13px;color:var(--text-secondary);padding:4px 12px;background:var(--bg-surface);border:1px solid var(--border);border-radius:20px;line-height:1.5}
+.meta-tag code{font-size:12px;color:var(--text-secondary);background:transparent;padding:0;letter-spacing:0.02em}
+
+/* Run ID stamp — H1 保留完整 run-id (与列表页一致), 时间戳从 ID 解出来作为副标小字,
+   告诉用户「这次跑的时间」,无需用户自己解析 ID 后缀。 */
+.run-id-stamp{font-size:0.75rem;color:var(--text-muted);margin:2px 0 12px;font-variant-numeric:tabular-nums}
+
+/* Audit fingerprints — judge prompt hash + 执行环境 fingerprint 默认折叠,
+   平日 review 不需要看, 复现 / 审计时再展开。 */
+.audit-fingerprints{margin:0 0 18px}
+.audit-fingerprints>summary{cursor:pointer;font-size:0.75rem;color:var(--text-muted);padding:4px 0;list-style:revert;user-select:none;transition:color 0.15s}
+.audit-fingerprints>summary:hover{color:var(--text-secondary)}
+.audit-fingerprints[open]>summary{color:var(--text-secondary);margin-bottom:6px}
+.audit-fingerprints[open]>.meta-tags{margin-top:0}
+
+/* Methodology audit — 方法学证据 (评委一致 / 差异显著 / 已饱和 / 人工对齐) 默认折叠,
+   summary 行直接显示 4 个 badge: 全 ✓ = 流程健康,任一 ⚠/✗ = 强制 open. */
+.methodology-audit{margin:24px 0;border:1px solid var(--border);border-radius:var(--radius);background:var(--bg-elevated)}
+.methodology-audit>summary{cursor:pointer;display:flex;flex-wrap:wrap;align-items:center;gap:10px 14px;padding:12px 16px;font-size:13px;color:var(--text-primary);font-weight:600;list-style:revert;user-select:none}
+.methodology-audit>summary:hover{color:var(--accent)}
+.methodology-summary-label{font-size:14px;font-weight:600}
+.methodology-summary-hint{flex:1 1 auto;text-align:right;font-size:11px;color:var(--text-muted);font-weight:400}
+.methodology-badges{display:inline-flex;flex-wrap:wrap;gap:6px}
+.methodology-badge{display:inline-flex;align-items:center;gap:4px;font-size:12px;font-weight:600;padding:3px 10px;border-radius:12px;letter-spacing:0.02em;line-height:1.5;white-space:nowrap}
+/* Pantone 哑色 — 用 var(--green-bg) 等(浅色主题已重定义) */
+.methodology-badge-pass{background:var(--green-bg);color:var(--green)}
+.methodology-badge-warn{background:var(--yellow-bg);color:var(--yellow)}
+.methodology-badge-fail{background:var(--red-bg);color:var(--red)}
+.methodology-badge-skip{background:var(--bg-soft);color:var(--text-muted)}
+.methodology-body{padding:0 16px 14px;border-top:1px solid var(--border);margin-top:8px}
+.methodology-body>h2:first-child{margin-top:14px}
+
+/* Cards */
+.cards{display:flex;gap:12px;flex-wrap:wrap;margin:10px 0}
+.card{background:var(--bg-surface);border:1px solid var(--border);border-radius:var(--radius);padding:14px 18px;min-width:140px;flex:1;transition:border-color 0.15s}
+.card:hover{border-color:var(--border-hover)}
+/* 卡片标签:de-uppercase(uppercase 在 pantone 哑色调里太"科技报"),轻微 letter-spacing
+   保留排版感,字号上调到 12 让读者看清。值/数字改 weight 600(700 在浅色配色里太硬)。 */
+.card-label{font-size:12px;color:var(--text-muted);letter-spacing:0.02em;margin-bottom:4px;font-weight:500}
+.card-value{font-size:20px;font-weight:600;margin:2px 0;color:var(--text-primary);font-variant-numeric:tabular-nums;line-height:1.3}
+.card-sub{font-size:12px;color:var(--text-muted);line-height:1.5}
+/* Summary table — inherits the global center + middle from the base td/th. */
+.summary-cell{min-width:96px}
+.summary-value-primary{font-size:1.25rem;font-weight:600;font-variant-numeric:tabular-nums}
+
+/* Hint tooltip (legacy span-based, kept for hover-only hints) */
+.hint{position:relative;display:inline-flex;align-items:center;justify-content:center;width:15px;height:15px;font-size:10px;font-weight:600;color:var(--text-muted);border:1px solid var(--border-hover);border-radius:50%;cursor:help;margin-left:6px;vertical-align:middle}
+.hint-click{cursor:pointer}
+
+/* Hint button — accessible, keyboard-focusable replacement for click-to-open-modal hints */
+button.hint-btn{display:inline-flex;align-items:center;justify-content:center;min-width:20px;min-height:20px;padding:2px;font-size:11px;font-weight:600;color:var(--text-muted);background:transparent;border:1px solid var(--border);border-radius:50%;cursor:pointer;margin-left:6px;vertical-align:middle;line-height:1;transition:color 0.15s,border-color 0.15s;outline:none;appearance:none}
+button.hint-btn:hover{color:var(--text-primary);border-color:var(--text-secondary)}
+button.hint-btn:focus-visible{outline:2px solid var(--accent);outline-offset:2px}
+button.hint-btn:focus-visible{outline:2px solid var(--accent);outline-offset:2px}
+
+/* Verdict + detail pattern used in variance & significance cells */
+.verdict-line{color:var(--text-secondary);font-size:var(--fs-body)}
+.detail-line{font-size:var(--fs-detail);color:var(--text-muted);margin-top:2px}
+
+/* Modal glossary layout — grid rows with tree connectors for sub-items */
+.modal-glossary-hint{font-size:var(--fs-detail);color:var(--text-muted);margin:4px 0 14px;font-style:italic}
+.modal-glossary{display:flex;flex-direction:column}
+.modal-glossary-row{display:grid;grid-template-columns:100px 1fr;gap:16px;padding:9px 0;border-bottom:1px solid var(--border)}
+.modal-glossary-row:last-child{border-bottom:none}
+.modal-glossary-label{font-size:var(--fs-body);color:var(--text-primary);font-weight:600}
+.modal-glossary-desc{font-size:var(--fs-detail);color:var(--text-secondary);line-height:1.55}
+.modal-glossary-sub{display:grid;grid-template-columns:86px 1fr;gap:14px;padding:6px 0 6px 28px;position:relative}
+.modal-glossary-sub::before{content:'';position:absolute;left:10px;top:0;bottom:0;width:2px;background:var(--border-hover);border-radius:1px}
+.modal-glossary-sub-label{font-size:var(--fs-detail);color:var(--text-secondary);font-weight:500}
+.modal-glossary-sub-desc{font-size:var(--fs-detail);color:var(--text-muted);line-height:1.55}
+
+/* Modal section divider */
+.modal-section{margin-top:20px;padding-top:16px;border-top:1px solid var(--border-hover)}
+.modal-section-title{font-size:var(--fs-body);font-weight:600;color:var(--text-primary);margin-bottom:10px}
+
+/* Four-quadrant diagnostic rule cards (matches the table's icon+text style) */
+.diag-rules{display:flex;flex-direction:column;gap:8px}
+.diag-rule-row{display:flex;align-items:flex-start;gap:10px;padding:10px 12px;background:var(--bg-elevated);border-radius:var(--radius);border-left:3px solid var(--border-hover)}
+.diag-rule-row.rule-good{border-left-color:var(--green)}
+.diag-rule-row.rule-warn{border-left-color:var(--yellow)}
+.diag-rule-row.rule-neutral{border-left-color:var(--text-muted)}
+.diag-rule-icon{font-size:15px;flex-shrink:0;line-height:1.4}
+.diag-rule-icon.rule-good{color:var(--green)}
+.diag-rule-icon.rule-warn{color:var(--yellow)}
+.diag-rule-icon.rule-neutral{color:var(--text-muted)}
+.diag-rule-body{flex:1;min-width:0}
+.diag-rule-title{font-size:var(--fs-detail);font-weight:600;color:var(--text-primary);margin-bottom:3px}
+.diag-rule-desc{font-size:var(--fs-detail);color:var(--text-secondary);line-height:1.5}
+.diag-rule-example{font-size:var(--fs-micro);color:var(--text-muted);font-family:ui-monospace,SFMono-Regular,Menlo,monospace;margin-top:4px;opacity:0.85}
+
+/* Variance & significance table: all cells use the global center + middle. */
+.variance-table td{padding-top:12px;padding-bottom:12px}
+.variance-table td.diagnostic-cell{min-width:180px}
+.variance-table .diag-faded strong{opacity:0.5;font-weight:500}
+
+/* Knowledge Interaction section (v0.17 / A):
+   variant card is the only visual container, two inner columns use a
+   single vertical divider instead of nested bg — compresses hierarchy
+   from 3-4 layers to 2. */
+.ki-card{margin-bottom:10px;padding:12px 14px;background:var(--bg-surface);border:1px solid var(--border);border-radius:var(--radius)}
+.ki-card-header{display:flex;flex-direction:column;gap:3px;margin-bottom:10px}
+.ki-card-title{font-size:15px;font-weight:600;color:var(--text-primary)}
+.ki-card-meta{font-size:var(--fs-micro);color:var(--text-muted);font-weight:400}
+.ki-columns{display:flex;gap:0;flex-wrap:wrap}
+.ki-col{flex:1;min-width:220px;padding:0 18px}
+.ki-col:first-child{padding-left:0;border-right:1px solid var(--border)}
+.ki-col:last-child{padding-right:0}
+@media(max-width:640px){
+  .ki-col{padding:0;min-width:100%;border-right:none !important}
+  .ki-col+.ki-col{margin-top:16px;padding-top:14px;border-top:1px solid var(--border)}
+}
+.ki-col-header{display:flex;justify-content:space-between;align-items:baseline;margin-bottom:8px}
+.ki-col-title{font-size:13px;color:var(--text-secondary);font-weight:500}
+.ki-col-value{font-size:22px;font-weight:600}
+.ki-bar{height:6px;background:var(--bg-card);border-radius:4px;margin-bottom:8px;overflow:hidden}
+.ki-bar-fill{height:100%;border-radius:4px;transition:width 0.2s}
+.ki-inventory-item{padding:6px 10px;margin:4px 0;background:var(--bg-card);border-left:3px solid var(--border-hover);border-radius:4px;font-size:var(--fs-detail);line-height:1.5}
+.ki-inventory-item[data-severity="strong"]{border-left-color:var(--red)}
+.ki-inventory-item[data-severity="medium"]{border-left-color:var(--yellow)}
+.ki-inventory-item[data-severity="weak"]{border-left-color:var(--text-muted)}
+.ki-inventory-item-meta{color:var(--text-muted);font-size:var(--fs-micro);margin-bottom:2px}
+.ki-inventory-item-ctx{color:var(--text-secondary);word-break:break-all}
+.ki-desc{font-size:12px;color:var(--text-muted);margin-bottom:4px;line-height:1.6}
+.ki-desc-hint{font-size:11px;color:var(--text-faint);margin-bottom:12px;line-height:1.5}
+.ki-details{margin-top:8px;padding-top:8px;border-top:1px solid var(--border)}
+.ki-details>summary{cursor:pointer;font-size:var(--fs-micro);color:var(--text-muted);padding:2px 0;list-style:revert;user-select:none;transition:color 0.15s}
+.ki-details>summary:hover{color:var(--text-secondary)}
+.ki-details[open]>summary{color:var(--text-secondary);margin-bottom:6px}
+
+/* Three-layer independent significance breakdown (PR-2).
+   Default collapsed; expands inline under each comparison. */
+.layer-breakdown-row>td{padding:0 !important;background:transparent;border-top:1px dashed var(--border-hover)}
+.layer-breakdown{padding:10px 16px 14px 32px;background:var(--bg-elevated)}
+.layer-breakdown>summary{cursor:pointer;font-size:var(--fs-detail);color:var(--text-muted);padding:4px 0;list-style:revert;user-select:none}
+.layer-breakdown>summary:hover{color:var(--text-secondary)}
+.layer-breakdown[open]>summary{color:var(--text-primary);margin-bottom:8px}
+.layer-sub-table{margin:0;width:100%;font-size:var(--fs-detail)}
+.layer-sub-table td{padding-top:8px;padding-bottom:8px}
+/* Multiple-comparisons disclaimer for the three-layer breakdown (PR-2) */
+.layer-breakdown-disclaimer{font-size:var(--fs-micro);color:var(--text-muted);line-height:1.5;padding:4px 8px 10px 0;font-style:italic}
+
+/* v0.21 B.4 — Verdict pill (GitHub PR check 风, 水平 status banner + CTA + 层 strip).
+   颜色严格按 level 编码: PROGRESS 绿, REGRESS 红, CAUTIOUS/UNDERPOWERED 黄,
+   NOISE/SOLO 用 secondary 而非 muted (深底下保证对比度). border-left 4px 是
+   status 视觉信号, 不堆边框/卡片 — 整体融入页面排版. */
+.verdict-banner{display:flex;align-items:baseline;gap:10px;flex-wrap:wrap;padding:12px 18px;margin:12px 0 0;border-left:4px solid var(--text-secondary);background:rgba(148,163,184,0.06);border-radius:var(--radius)}
+.verdict-banner.verdict-PROGRESS{border-left-color:var(--green);background:rgba(74,222,128,0.07)}
+.verdict-banner.verdict-REGRESS{border-left-color:var(--red);background:rgba(248,113,113,0.07)}
+.verdict-banner.verdict-CAUTIOUS{border-left-color:var(--yellow);background:rgba(251,191,36,0.07)}
+.verdict-banner.verdict-UNDERPOWERED{border-left-color:var(--yellow);background:rgba(251,191,36,0.04)}
+.verdict-banner.verdict-NOISE{border-left-color:var(--text-secondary);background:rgba(148,163,184,0.06)}
+.verdict-banner.verdict-SOLO{border-left-color:var(--border-hover);background:transparent}
+.verdict-icon{font-size:12px;line-height:1;flex-shrink:0}
+.verdict-line{font-size:14px;font-weight:500;color:var(--text-primary);line-height:1.6;letter-spacing:-0.005em}
+
+/* page-verdict: verdict 是报告的「答案」,头部 hero 形式呈现.
+   行 1: enum + ship-action badge (PROGRESS · SHIP) + 自然语言 ("明显更好,可以发布")
+   行 2: Δ / 95% CI / N / CV — 测量学核心数字, 不需要滚到下方表才看到.
+   颜色按 level 编码 (PROGRESS 绿 / REGRESS 红 / CAUTIOUS+UNDERPOWERED 黄 / NOISE+SOLO 中性). */
+.page-verdict{display:flex;flex-direction:column;gap:10px;margin:12px 0 18px;padding:14px 18px;border-left:4px solid var(--text-secondary);background:rgba(148,163,184,0.06);border-radius:var(--radius)}
+.page-verdict.verdict-PROGRESS{border-left-color:var(--green);background:rgba(74,222,128,0.07)}
+.page-verdict.verdict-REGRESS{border-left-color:var(--red);background:rgba(248,113,113,0.07)}
+.page-verdict.verdict-CAUTIOUS{border-left-color:var(--yellow);background:rgba(251,191,36,0.07)}
+.page-verdict.verdict-UNDERPOWERED{border-left-color:var(--yellow);background:rgba(251,191,36,0.04)}
+.page-verdict.verdict-NOISE{border-left-color:var(--text-secondary);background:rgba(148,163,184,0.06)}
+.page-verdict.verdict-SOLO{border-left-color:var(--border-hover);background:transparent}
+.page-verdict-head{display:flex;align-items:baseline;gap:12px;flex-wrap:wrap}
+.page-verdict-badge{display:inline-flex;align-items:center;gap:6px;font-family:ui-monospace,SFMono-Regular,Menlo,monospace;font-size:12px;font-weight:700;letter-spacing:0.04em;padding:3px 10px;border-radius:14px;background:rgba(148,163,184,0.14);color:var(--text-primary);white-space:nowrap}
+.verdict-PROGRESS .page-verdict-badge{background:rgba(74,222,128,0.18);color:var(--green)}
+.verdict-REGRESS .page-verdict-badge{background:rgba(248,113,113,0.18);color:var(--red)}
+.verdict-CAUTIOUS .page-verdict-badge,
+.verdict-UNDERPOWERED .page-verdict-badge{background:rgba(251,191,36,0.18);color:var(--yellow)}
+.verdict-NOISE .page-verdict-badge,
+.verdict-SOLO .page-verdict-badge{background:rgba(148,163,184,0.16);color:var(--text-secondary)}
+.page-verdict-badge-dot{font-size:8px;line-height:1}
+.page-verdict-text{font-size:15px;font-weight:500;color:var(--text-primary);line-height:1.5;flex:1 1 auto;min-width:0}
+.page-verdict-metrics{display:flex;flex-wrap:wrap;gap:10px 20px;align-items:baseline;padding-top:2px}
+.verdict-metric{display:inline-flex;align-items:baseline;gap:6px}
+.verdict-metric-label{font-size:11px;color:var(--text-secondary);text-transform:uppercase;letter-spacing:0.06em;font-weight:600}
+.verdict-metric-value{font-size:15px;font-weight:700;color:var(--text-primary);font-variant-numeric:tabular-nums;letter-spacing:-0.01em}
+
+.verdict-cta{display:flex;align-items:baseline;gap:8px;flex-wrap:wrap;padding:10px 18px;margin:0;border-left:4px solid var(--accent);background:var(--bg-surface);border-top:1px solid var(--border);border-radius:0}
+.verdict-cta.verdict-cta-PROGRESS{border-left-color:var(--green);background:rgba(74,222,128,0.04)}
+.verdict-cta.verdict-cta-REGRESS{border-left-color:var(--red);background:rgba(248,113,113,0.04)}
+.verdict-cta.verdict-cta-CAUTIOUS{border-left-color:var(--yellow);background:rgba(251,191,36,0.04)}
+.verdict-cta.verdict-cta-NOISE,
+.verdict-cta.verdict-cta-UNDERPOWERED,
+.verdict-cta.verdict-cta-SOLO{border-left-color:var(--text-secondary)}
+.verdict-cta-icon{font-size:13px;flex-shrink:0;color:var(--text-secondary)}
+.verdict-cta-PROGRESS .verdict-cta-icon{color:var(--green)}
+.verdict-cta-REGRESS .verdict-cta-icon{color:var(--red)}
+.verdict-cta-CAUTIOUS .verdict-cta-icon{color:var(--yellow)}
+.verdict-cta-action{font-size:13px;font-weight:700;color:var(--text-primary);letter-spacing:0.04em;text-transform:uppercase;white-space:nowrap}
+.verdict-cta-detail{font-size:13px;color:var(--text-secondary);line-height:1.55;flex:1 1 200px;min-width:0}
+
+.verdict-layers{display:flex;align-items:baseline;gap:10px;padding:6px 18px 0;margin:0 0 16px;border-left:4px solid transparent;flex-wrap:wrap}
+.verdict-layers-label{font-size:10px;text-transform:uppercase;letter-spacing:0.1em;color:var(--text-muted);font-weight:500}
+.verdict-layers-content{font-size:12px;color:var(--text-secondary);font-variant-numeric:tabular-nums}
+
+/* v0.21 B.4 — 列表页 RUN ID 旁的 verdict status pill. inline-flex 紧凑 pill,
+   颜色编码同 banner, 字号微小但对比清晰. 鼠标悬停/键盘聚焦时不抢戏 — 它只是
+   "这个 run 是什么 status" 的一眼瞥. */
+.run-status{display:inline-flex;align-items:center;gap:4px;font-size:11px;font-weight:600;padding:1px 8px;border-radius:var(--radius);margin-right:8px;line-height:1.5;letter-spacing:-0.005em;white-space:nowrap;color:var(--text-secondary);background:rgba(148,163,184,0.1);vertical-align:1px}
+.run-status .run-status-dot{font-size:8px;line-height:1}
+.run-status.verdict-PROGRESS{color:var(--green);background:rgba(74,222,128,0.12)}
+.run-status.verdict-REGRESS{color:var(--red);background:rgba(248,113,113,0.12)}
+.run-status.verdict-CAUTIOUS{color:var(--yellow);background:rgba(251,191,36,0.12)}
+.run-status.verdict-UNDERPOWERED{color:var(--yellow);background:rgba(251,191,36,0.08)}
+.run-status.verdict-NOISE{color:var(--text-secondary);background:rgba(148,163,184,0.1)}
+.run-status.verdict-SOLO{color:var(--text-muted);background:transparent;border:1px solid var(--border)}
+
+@media print{
+  .verdict-banner,.verdict-cta{break-inside:avoid;page-break-inside:avoid}
+  .run-status{border:1px solid #cbd5e1;background:transparent !important}
+}
+
+.modal-overlay{display:none;position:fixed;inset:0;z-index:999;background:rgba(0,0,0,0.6);align-items:center;justify-content:center}
+.modal-content{background:var(--bg-card);border:1px solid var(--border);border-radius:var(--radius);max-width:800px;max-height:80vh;overflow:auto;padding:24px;margin:20px;width:90%}
+.modal-header{display:flex;justify-content:space-between;align-items:center;margin-bottom:8px}
+.modal-close{cursor:pointer;background:none;border:none;color:var(--text-muted);font-size:18px;padding:8px 12px;border-radius:var(--radius);transition:background 0.15s,color 0.15s}
+.modal-close:hover{color:var(--text-primary);background:var(--bg-surface)}
+.modal-table{width:100%;font-size:13px;margin:12px 0;background:transparent;border:none;table-layout:auto}
+.modal-table td{padding:6px 0;border:none;background:transparent;word-break:break-word;overflow-wrap:anywhere}
+.modal-table td:first-child{white-space:nowrap;vertical-align:top;min-width:80px;padding-right:16px;word-break:keep-all}
+/* Inline <code> inside modal text — improve readability on dark surfaces */
+.modal-table code,.modal-glossary code,.modal-section code{background:var(--bg-surface);padding:1px 6px;border-radius:3px;font-size:var(--fs-micro);color:var(--text-primary);font-family:ui-monospace,SFMono-Regular,Menlo,monospace}
+@media(max-width:480px){.modal-table td{display:block;padding:3px 0}.modal-table td:first-child{font-weight:600}}
+.hint-tip{display:none;position:absolute;bottom:calc(100% + 6px);right:0;background:var(--bg-elevated);border:1px solid var(--border-hover);border-radius:var(--radius);padding:6px 10px;font-size:11px;font-weight:400;color:var(--text-secondary);white-space:normal;max-width:280px;width:max-content;z-index:10}
+.hint:hover .hint-tip,.hint:focus .hint-tip{display:block}
+.summary-value{font-size:15px;font-weight:600;color:var(--text-primary);font-variant-numeric:tabular-nums}
+.summary-detail{font-size:12px;color:var(--text-muted);margin-top:4px;line-height:1.5}
+.summary-unit{font-size:12px;font-weight:400;color:var(--text-muted)}
+.card-detail{margin-top:8px;font-size:12px;color:var(--text-secondary)}
+.card-detail div{margin:2px 0}
+
+/* Table */
+.table-wrap{overflow-x:auto;-webkit-overflow-scrolling:touch;margin:16px 0;position:relative}
+/* 窄屏下 6 列宽 >700px 会横向溢出,右侧稳定性列容易被遮盖。
+   加一条渐变阴影作为"可滑动"视觉提示,仅在 ≤768px 且可滚动容器里显示。 */
+@media(max-width:768px){
+  .table-wrap::after{content:'';position:sticky;right:0;top:0;display:block;float:right;width:32px;height:100%;margin-left:-32px;margin-top:-100%;pointer-events:none;background:linear-gradient(to right,transparent,var(--bg-card) 85%);z-index:2}
+}
+/* 表格 — 稍紧凑(padding 7px 12px),正文 13px,字号上调跟全站对齐;
+   非数字列保留 left-align(td.text 类),其它默认居中。 */
+table{border-collapse:collapse;width:100%;font-size:13px;background:var(--bg-surface);border:1px solid var(--border);border-radius:var(--radius);overflow:hidden;line-height:1.5}
+th{background:var(--bg-elevated);padding:8px 12px;text-align:center;vertical-align:middle;font-weight:600;color:var(--text-secondary);border-bottom:1px solid var(--border);font-size:12px;letter-spacing:0.02em;white-space:nowrap}
+td{padding:7px 12px;border-bottom:1px solid var(--border);color:var(--text-secondary);font-variant-numeric:tabular-nums;text-align:center;vertical-align:middle}
+td:first-child, th:first-child { text-align:left }
+tr:last-child td{border-bottom:none}
+tr:hover td{background:rgba(148,163,184,0.04)}
+
+/* Badges */
+.badge{display:inline-block;padding:2px 8px;border-radius:var(--radius);font-size:11px;font-weight:600}
+.badge-ok{background:var(--green-bg);color:var(--green)}
+.badge-err{background:var(--red-bg);color:var(--red)}
+.badge-pass{background:var(--green-bg);color:var(--green)}
+.badge-fail{background:var(--red-bg);color:var(--red)}
+.badge-muted{background:var(--bg-surface);color:var(--text-muted);border:1px solid var(--border)}
+
+/* Nav */
+.nav{margin-bottom:24px;font-size:13px}
+
+/* Error detail */
+.error-detail{display:inline-block;font-size:11px;color:var(--red);word-break:break-all;max-width:260px}
+
+/* Assertion & dimension tags */
+.assertion-list{margin:4px 0;padding:0;list-style:none;font-size:11px}
+.assertion-list li{margin:2px 0;color:var(--text-secondary)}
+.dim-scores{display:flex;gap:6px;flex-wrap:wrap;margin-top:4px}
+.dim-tag{font-size:10px;padding:2px 7px;border-radius:4px;background:var(--info-bg);color:var(--accent)}
+.dim-desc{font-size:11px;color:var(--text-muted);font-weight:400;margin-left:6px}
+
+/* Bar chart */
+.bar-row{display:flex;align-items:center;gap:8px;margin:6px 0}
+.bar-label{flex:0 0 60px;font-size:12px;color:var(--text-muted)}
+.bar-fill{flex:1;height:16px;border-radius:4px;opacity:0.8}
+.bar-value{flex:0 0 auto;font-size:12px;color:var(--text-secondary)}
+
+/* Forms */
+input[type="text"]{background:var(--bg-surface);border:1px solid var(--border);color:var(--text-primary);border-radius:var(--radius);padding:6px 12px;font-size:13px;font-family:inherit;outline:none;transition:border-color .15s,box-shadow .15s}
+input[type="text"]:focus{border-color:var(--accent);box-shadow:0 0 0 2px var(--info-bg)}
+/* 通用 button — 浅色主题里默认 transparent + 边框更细。子样式(.run-card-delete, .tv-vtab,
+   .omk-view-tab, .hint-btn 等)各自有显式 reset 覆盖默认样式。 */
+button{background:var(--bg-surface);border:1px solid var(--border);color:var(--text-secondary);border-radius:var(--radius);cursor:pointer;padding:6px 14px;font-size:13px;font-family:inherit;transition:border-color .15s,color .15s,background .15s;outline:none;appearance:none;-webkit-appearance:none}
+button:hover{border-color:var(--border-hover);color:var(--text-primary)}
+button:focus-visible{outline:2px solid var(--accent);outline-offset:2px}
+
+/* Lang toggle */
+.lang-toggle{position:fixed;top:16px;right:16px;padding:6px 14px;z-index:100;font-size:12px}
+
+/* Misc */
+.btn-danger{color:var(--red)}
+
+/* Focus */
+a:focus-visible,.badge:focus-visible{outline:2px solid var(--accent);outline-offset:2px;border-radius:4px}
+
+/* Responsive */
+@media(max-width:768px){
+  body{padding:16px}
+  .cards{flex-direction:column}
+  .card{min-width:0}
+  h1{font-size:1.25rem}
+  h2{font-size:0.875rem}
+  .lang-toggle{top:8px;right:8px;padding:4px 10px;font-size:11px}
+  button{min-height:44px;padding:8px 14px}
+  /* Override: hint buttons stay compact so they don't dominate mobile layout */
+  button.hint-btn{min-width:28px;min-height:28px;padding:4px}
+  /* Bump detail font so secondary lines are readable on mobile */
+  .detail-line{font-size:var(--fs-label)}
+}
+@media(max-width:480px){
+  body{padding:12px}
+  td,th{padding:8px 10px;font-size:12px}
+}
+@media print{
+  body{background:#fff;color:#1e293b;padding:20px;max-width:none}
+  h1,h2,.card-value,.summary-value,.summary-value-primary{color:#1e293b}
+  .card,.summary-table,table{background:#fff;border-color:#e2e8f0}
+  th{background:#f8fafc;color:#475569;border-color:#e2e8f0}
+  td{color:#334155;border-color:#f1f5f9}
+  .badge-ok,.badge-pass{background:#dcfce7;color:#166534}
+  .badge-err,.badge-fail{background:#fee2e2;color:#991b1b}
+  .badge-muted{background:#f8fafc;color:#94a3b8;border-color:#e2e8f0}
+  .dim-tag{background:#f1f5f9;color:#3b82f6}
+  .meta-tag{background:#f8fafc;border-color:#e2e8f0;color:#475569}
+  .bar-fill{opacity:1}
+  a{color:#1e293b;text-decoration:none}
+  .lang-toggle,.btn-danger,.nav{display:none}
+  .hint-tip{display:none}
+  .footer{color:#475569}
+}
+</style></head><body>${body}<footer class="footer" style="margin-top:40px;padding-top:16px;border-top:1px solid var(--border);font-size:11px;color:var(--text-faint);text-align:center">Powered by oh-my-knowledge</footer>${langToggleScript()}</body></html>`;
+}
+//# sourceMappingURL=layout.js.map