oh-my-knowledge 0.28.0 → 0.30.0

package/dist/src/renderer/html-renderer.js

@@ -2,8 +2,9 @@
  * HTML report renderer — orchestrates sub-modules.
  */
 import { e, fmtCost, fmtDuration, fmtKnownCost, COLORS, DEFAULT_LANG, t, layout } from './layout.js';
-import { renderAgentOverview, renderAnalysis, renderHumanAgreement, renderKnowledgeInteractionSection, renderPairwiseDiff, renderSaturationCurve, renderSummaryCards, renderVarianceComparisons, renderVerdictPill, levelLabel, levelTooltip, } from './summary.js';
+import { renderAgentOverview, renderAnalysis, renderKnowledgeInteractionSection, renderMethodologyAudit, renderScoringModal, renderSummaryCards, renderVarianceComparisons, renderVerdictPill, levelLabel, levelTooltip, } from './summary.js';
 import { renderSampleTable } from './table.js';
+import { renderTestView, TEST_VIEW_CSS, TEST_VIEW_JS } from './test-view.js';
 import { renderTrendsBody } from './trends.js';
 import { computeVerdict } from '../eval-core/verdict.js';
 // v0.21 B.4 — 列表页 status pill 用的 dot. PROGRESS/REGRESS 实心(强信号),
@@ -31,36 +32,6 @@ function improvementOf(baselineScore, skillScore) {
     const delta = ((skillScore - baselineScore) / baselineScore * 100).toFixed(0);
     return skillScore >= baselineScore ? `+${delta}%` : `${delta}%`;
 }
-function renderRuntimeTag(runtime, label, lang) {
-    if (!runtime)
-        return '';
-    const versions = [];
-    if (runtime.binary?.version)
-        versions.push(`${lang === 'zh' ? '二进制' : 'binary'} ${runtime.binary.version}`);
-    if (runtime.sdk?.version)
-        versions.push(`sdk ${runtime.sdk.version}`);
-    const versionText = versions.length > 0 ? ` · ${versions.map(e).join(' · ')}` : '';
-    const title = lang === 'zh'
-        ? [
-            `executor=${runtime.executor}`,
-            `model=${runtime.model}`,
-            `kind=${runtime.kind}`,
-            `system=${runtime.capabilities.systemPrompt}`,
-            `cost=${runtime.capabilities.costUSD}`,
-            `trace=${runtime.capabilities.trace}`,
-            `skillIsolation=${runtime.capabilities.skillIsolation}`,
-        ].join('; ')
-        : [
-            `executor=${runtime.executor}`,
-            `model=${runtime.model}`,
-            `kind=${runtime.kind}`,
-            `system=${runtime.capabilities.systemPrompt}`,
-            `cost=${runtime.capabilities.costUSD}`,
-            `trace=${runtime.capabilities.trace}`,
-            `skillIsolation=${runtime.capabilities.skillIsolation}`,
-        ].join('; ');
-    return `<span class="meta-tag" title="${e(title)}">${e(label)}: <code>${e(runtime.fingerprint)}</code>${versionText}</span>`;
-}
 function costCompletenessTooltip(lang) {
     return lang === 'zh'
         ? '部分 executor 或评委不回传 USD 成本。这里显示的是已上报成本下界,真实花费可能更高。'
@@ -84,33 +55,85 @@ function renderDebiasModeTag(modes, lang) {
 function pluralizeEn(count, singular, plural = `${singular}s`) {
     return `${count} ${count === 1 ? singular : plural}`;
 }
-function renderJudgeRuntimeTags(meta, lang) {
-    if (meta.noJudge)
-        return '';
-    const entries = meta.judgeModels ?? [];
-    if (entries.length === 0)
-        return '';
-    if (entries.length === 1) {
-        return renderRuntimeTag(entries[0].runtime, lang === 'zh' ? '评委指纹' : 'Judge runtime', lang);
-    }
-    return entries
-        .slice()
-        .sort((a, b) => `${a.executor}:${a.model}`.localeCompare(`${b.executor}:${b.model}`))
-        .map((entry) => renderRuntimeTag(entry.runtime, lang === 'zh' ? `评委指纹 ${entry.executor}:${entry.model}` : `Judge runtime ${entry.executor}:${entry.model}`, lang))
-        .join('');
-}
-function renderExecutorRuntimeTags(meta, lang) {
+function gatherRuntimeScopes(meta) {
+    const scopes = [];
     if (meta.executorRuntimes && Object.keys(meta.executorRuntimes).length > 0) {
-        return Object.entries(meta.executorRuntimes)
+        Object.entries(meta.executorRuntimes)
             .sort(([a], [b]) => a.localeCompare(b))
-            .map(([key, runtime]) => renderRuntimeTag(runtime, lang === 'zh' ? `执行器指纹 ${key}` : `Executor runtime ${key}`, lang))
-            .join('');
+            .forEach(([key, runtime]) => {
+            if (runtime)
+                scopes.push({ role: 'executor', scope: key, runtime });
+        });
+    }
+    else if (meta.executorRuntime) {
+        scopes.push({ role: 'executor', scope: '', runtime: meta.executorRuntime });
+    }
+    if (!meta.noJudge) {
+        (meta.judgeModels ?? [])
+            .slice()
+            .sort((a, b) => `${a.executor}:${a.model}`.localeCompare(`${b.executor}:${b.model}`))
+            .forEach((entry) => {
+            if (entry.runtime)
+                scopes.push({ role: 'judge', scope: `${entry.executor}:${entry.model}`, runtime: entry.runtime });
+        });
+    }
+    return scopes;
+}
+function runtimeVersionText(runtime, lang) {
+    const versions = [];
+    if (runtime.binary?.version)
+        versions.push(`${lang === 'zh' ? '二进制' : 'binary'} ${runtime.binary.version}`);
+    if (runtime.sdk?.version)
+        versions.push(`sdk ${runtime.sdk.version}`);
+    return versions.length > 0 ? ` · ${versions.map(e).join(' · ')}` : '';
+}
+function runtimeTooltip(runtime) {
+    return [
+        `executor=${runtime.executor}`,
+        `model=${runtime.model}`,
+        `kind=${runtime.kind}`,
+        `system=${runtime.capabilities.systemPrompt}`,
+        `cost=${runtime.capabilities.costUSD}`,
+        `trace=${runtime.capabilities.trace}`,
+        `skillIsolation=${runtime.capabilities.skillIsolation}`,
+    ].join('; ');
+}
+// 同 fingerprint + 同 binary/sdk 版本 = 同一执行环境，合并为一条 tag。
+// 旧版给每个 executor / judge 单独一条 pill，3 个角色用同一 binary 时
+// fingerprint 重复 3 遍，扫读成本高。新版按 (fingerprint, versionText)
+// 分组，scope 合并到 tag 内 "适用于 ..." 后缀。
+function renderRuntimeFingerprintTags(meta, lang) {
+    const scopes = gatherRuntimeScopes(meta);
+    if (scopes.length === 0)
+        return '';
+    const groups = new Map();
+    for (const s of scopes) {
+        const versionText = runtimeVersionText(s.runtime, lang);
+        const key = `${s.runtime.fingerprint}|${versionText}`;
+        let g = groups.get(key);
+        if (!g) {
+            g = { runtime: s.runtime, versionText, executors: [], judges: [] };
+            groups.set(key, g);
+        }
+        if (s.role === 'executor')
+            g.executors.push(s.scope || (lang === 'zh' ? '默认' : 'default'));
+        else
+            g.judges.push(s.scope);
     }
-    return renderRuntimeTag(meta.executorRuntime, lang === 'zh' ? '执行器指纹' : 'Executor runtime', lang);
+    const groupLabel = lang === 'zh' ? '执行环境指纹' : 'Runtime fingerprint';
+    return Array.from(groups.values()).map((g) => {
+        const scopeBits = [];
+        if (g.executors.length > 0)
+            scopeBits.push(`${lang === 'zh' ? '执行器' : 'executor'} ${g.executors.join(', ')}`);
+        if (g.judges.length > 0)
+            scopeBits.push(`${lang === 'zh' ? '评委' : 'judge'} ${g.judges.join(', ')}`);
+        const scopeText = scopeBits.length > 0 ? ` · ${lang === 'zh' ? '适用' : 'used by'} ${e(scopeBits.join('; '))}` : '';
+        return `<span class="meta-tag" title="${e(runtimeTooltip(g.runtime))}">${e(groupLabel)}: <code>${e(g.runtime.fingerprint)}</code>${g.versionText}${scopeText}</span>`;
+    }).join('');
 }
 export function renderRunList(runs, lang = DEFAULT_LANG) {
     const langQ = lang === DEFAULT_LANG ? '' : `?lang=${lang}`;
-    const skillHealthLink = `<a href="/analyses${langQ}" style="color:var(--text-muted);font-size:12px;text-decoration:none;border:1px solid var(--border);padding:4px 10px;border-radius:var(--radius);display:inline-block">📊 <span data-i18n="skillHealthTitle">${t('skillHealthTitle', lang)}</span> →</a>`;
+    const skillHealthLink = `<a class="page-secondary-link" href="/analyses${langQ}">📊 <span data-i18n="skillHealthTitle">${t('skillHealthTitle', lang)}</span> →</a>`;
     if (!runs || runs.length === 0) {
         return layout(t('title', lang), `
       <main>
@@ -121,112 +144,85 @@ export function renderRunList(runs, lang = DEFAULT_LANG) {
       </main>
     `, lang);
     }
-    const rows = runs.map((run) => {
+    // v0.30 — 卡片式列表(取代 7 列 dense table),每条占 4-5 行垂直空间但信息密度更可读。
+    // 顶部 verdict pill 主导视觉,id+date 是身份,scores 是成绩,底部 meta 是元数据,
+    // delete 默认隐藏在 hover 出现避免误点。批量报告共用同一组件,batch pill 替代 verdict。
+    const formatDateFromId = (id, fallbackTs) => {
+        const idMatch = id.match(/(\d{4})(\d{2})(\d{2})-(\d{2})(\d{2})$/);
+        if (idMatch)
+            return `${idMatch[2]}/${idMatch[3]} ${idMatch[4]}:${idMatch[5]}`;
+        return fallbackTs ? new Date(fallbackTs).toLocaleString('zh-CN', { month: '2-digit', day: '2-digit', hour: '2-digit', minute: '2-digit' }) : '';
+    };
+    const renderScoreBar = (label, score) => {
+        if (score == null)
+            return `<span class="rc-score-empty">${e(label)}: —</span>`;
+        const cls = score >= 4 ? 'pass' : score >= 3 ? 'warn' : 'fail';
+        const barW = Math.round((score / 5) * 100);
+        return `<div class="rc-score-row"><span class="rc-score-label" title="${e(label)}">${e(label)}</span>` +
+            `<div class="rc-score-bar"><div class="rc-score-bar-fill rc-score-bar-fill--${cls}" style="width:${barW}%"></div></div>` +
+            `<span class="rc-score-num rc-score-num--${cls}">${score.toFixed(2)}</span></div>`;
+    };
+    const cards = runs.map((run) => {
         if (run.kind === 'batch-evaluation') {
             const m = run.meta;
-            const scoreCol = run.items.length > 0
+            const scores = run.items.length > 0
                 ? run.items.map((item) => {
                     const baselineScore = scoreOf(item.summary.baseline);
                     const skillScore = scoreOf(item.summary[item.name]);
-                    const score = skillScore ?? baselineScore;
-                    if (score == null)
-                        return `<span style="color:var(--text-muted)">${e(item.name)}: -</span>`;
-                    const color = score >= 4 ? 'var(--green)' : score >= 3 ? 'var(--yellow)' : 'var(--red)';
-                    const barW = Math.round((score / 5) * 100);
-                    return `<div style="display:flex;align-items:center;gap:6px;margin:2px 0">` +
-                        `<span title="${e(item.name)}" style="font-size:11px;color:var(--text-muted);width:56px;text-align:right;overflow:hidden;text-overflow:ellipsis;white-space:nowrap;flex-shrink:0">${e(item.name)}</span>` +
-                        `<div style="width:64px;height:6px;background:var(--bg-surface);border-radius:3px;flex-shrink:0">` +
-                        `<div style="width:${barW}%;height:100%;background:${color};border-radius:3px"></div></div>` +
-                        `<span style="font-size:12px;font-weight:600;color:${color};min-width:24px">${score.toFixed(2)}</span></div>`;
+                    return renderScoreBar(item.name, skillScore ?? baselineScore);
                 }).join('')
-                : '<div style="color:var(--text-faint);font-size:0.6875rem;text-align:center">no score</div>';
-            const allCostReported = run.items.every((item) => Object.values(item.summary || {}).every((v) => v.execCostReported !== false && v.judgeCostReported !== false));
-            const totalCostReported = m.totalCostReported !== false && allCostReported;
+                : `<div class="rc-score-empty">${lang === 'zh' ? '无分数' : 'no score'}</div>`;
             const totalDurationMs = run.items.reduce((sum, item) => (sum + Object.values(item.summary || {}).reduce((inner, v) => inner + (v.avgDurationMs || 0) * (v.successCount || 0), 0)), 0);
             const statusPill = `<span class="run-status" title="${lang === 'zh' ? '批量评测' : 'batch evaluation'}"><span class="run-status-dot" aria-hidden="true">◇</span>${lang === 'zh' ? '批量' : 'Batch'}</span>`;
-            return `<tr>
-      <td>${statusPill}<a href="/reports/${e(run.id)}${langQ}"><span style="color:var(--text-primary)">${e(run.id)}</span><br><span style="font-size:0.6875rem;color:var(--text-muted)">${m.timestamp ? new Date(m.timestamp).toLocaleString('zh-CN', { month: '2-digit', day: '2-digit', hour: '2-digit', minute: '2-digit' }) : e(run.id)}</span></a></td>
-      <td>${e(m.model || '-')}</td>
-      <td>${m.sampleCount || 0}</td>
-      <td>${scoreCol}</td>
-      <td title="${totalCostReported ? '' : e(costCompletenessTooltip(lang))}">${fmtKnownCost(m.totalCostUSD || 0, totalCostReported)}</td>
-      <td>${fmtDuration(totalDurationMs)}</td>
-      <td style="white-space:nowrap"><button onclick="deleteRun('${e(run.id)}',this)" class="btn-danger" data-i18n="deleteBtnText">${t('deleteBtnText', lang)}</button></td>
-    </tr>`;
+            const date = formatDateFromId(run.id, m.timestamp);
+            return `<a class="run-card" href="/reports/${e(run.id)}${langQ}">
+        <div class="run-card-h">
+          ${statusPill}
+          <span class="run-card-id">${e(run.id)}</span>
+          ${date ? `<span class="run-card-date">${e(date)}</span>` : ''}
+        </div>
+        <div class="run-card-meta">${e(m.model || '-')} · ${m.sampleCount || 0} ${lang === 'zh' ? '用例' : 'samples'} · ${run.items.length} skills</div>
+        <div class="run-card-scores">${scores}</div>
+        <div class="run-card-foot">
+          <span class="run-card-stat">${fmtDuration(totalDurationMs)}</span>
+          <button class="run-card-delete" onclick="event.preventDefault();event.stopPropagation();deleteRun('${e(run.id)}',this)" data-i18n="deleteBtnText" aria-label="${e(t('deleteBtnText', lang))}">×</button>
+        </div>
+      </a>`;
         }
         const m = run.meta;
-        const hasScores = Object.values(run.summary || {}).some((s) => typeof s.avgCompositeScore === 'number' || typeof s.avgLlmScore === 'number');
-        const scoreCol = hasScores
-            ? Object.entries(run.summary || {}).map(([v, s]) => {
-                const score = s.avgCompositeScore ?? s.avgLlmScore ?? null;
-                if (score == null)
-                    return `<span style="color:var(--text-muted)">${e(v)}: -</span>`;
-                const color = score >= 4 ? 'var(--green)' : score >= 3 ? 'var(--yellow)' : 'var(--red)';
-                const barW = Math.round((score / 5) * 100);
-                return `<div style="display:flex;align-items:center;gap:6px;margin:2px 0">` +
-                    `<span title="${e(v)}" style="font-size:11px;color:var(--text-muted);width:56px;text-align:right;overflow:hidden;text-overflow:ellipsis;white-space:nowrap;flex-shrink:0">${e(v)}</span>` +
-                    `<div style="width:64px;height:6px;background:var(--bg-surface);border-radius:3px;flex-shrink:0">` +
-                    `<div style="width:${barW}%;height:100%;background:${color};border-radius:3px"></div></div>` +
-                    `<span style="font-size:12px;font-weight:600;color:${color};min-width:24px">${score}</span></div>`;
-            }).join('')
-            : '<div style="color:var(--text-faint);font-size:0.6875rem;text-align:center">no score</div>';
-        const badges = ''; // TODO: artifact kind 体系完善后，按 kind 显示评测类型标签
-        // v0.21 B.4 — 列表页 verdict pill: 一眼分辨 progress / regress / noise.
-        // computeVerdict 是同步纯函数(report -> level),per row 跑成本 O(samples).
-        // 脏报告可能让 layer-gates 访问 undefined.avgFactScore 抛 NPE。
-        // try/catch 兜底,失败的 row 不显示 pill,避免一个坏 report 撤掉整页。
+        const summary = run.summary || {};
+        const hasScores = Object.values(summary).some((s) => typeof s.avgCompositeScore === 'number' || typeof s.avgLlmScore === 'number');
+        const scoresHtml = hasScores
+            ? Object.entries(summary).map(([v, s]) => renderScoreBar(v, s.avgCompositeScore ?? s.avgLlmScore ?? null)).join('')
+            : `<div class="rc-score-empty">${lang === 'zh' ? '无分数' : 'no score'}</div>`;
         let statusPill = '';
+        let verdictLvl = '';
         try {
             const verdict = computeVerdict(run);
-            const lvl = verdict.level;
-            statusPill = `<span class="run-status verdict-${lvl}" title="${e(levelTooltip(lvl, lang))}" aria-label="${e(levelLabel(lvl, lang))} — ${e(levelTooltip(lvl, lang))}"><span class="run-status-dot" aria-hidden="true">${levelDot(lvl)}</span>${e(levelLabel(lvl, lang))}</span>`;
+            verdictLvl = verdict.level;
+            statusPill = `<span class="run-status verdict-${verdictLvl}" title="${e(levelTooltip(verdictLvl, lang))}" aria-label="${e(levelLabel(verdictLvl, lang))} — ${e(levelTooltip(verdictLvl, lang))}"><span class="run-status-dot" aria-hidden="true">${levelDot(verdictLvl)}</span>${e(levelLabel(verdictLvl, lang))}</span>`;
         }
         catch { /* skip pill on this row */ }
-        return `<tr>
-      <td>${statusPill}<a href="/reports/${e(run.id)}${langQ}"><span style="color:var(--text-primary)">${e(run.id)}${badges}</span><br><span style="font-size:0.6875rem;color:var(--text-muted)">${(() => {
-            // Extract date/time from report ID: ...-YYYYMMDD-HHmm
-            const idMatch = run.id.match(/(\d{4})(\d{2})(\d{2})-(\d{2})(\d{2})$/);
-            if (idMatch)
-                return `${idMatch[2]}/${idMatch[3]} ${idMatch[4]}:${idMatch[5]}`;
-            return m.timestamp ? new Date(m.timestamp).toLocaleString('zh-CN', { month: '2-digit', day: '2-digit', hour: '2-digit', minute: '2-digit' }) : e(run.id);
-        })()}</span></a></td>
-      <td>${e(m.model || '-')}</td>
-      <td>${m.sampleCount || 0}</td>
-      <td>${scoreCol}</td>
-      <td>${(() => {
-            const cost = Object.values(run.summary || {}).reduce((s, v) => s + (v.totalExecCostUSD || 0), 0);
-            const reported = Object.values(run.summary || {}).every((v) => v.execCostReported !== false);
-            return fmtCost(cost, reported);
-        })()}</td>
-      <td>${fmtDuration(Object.values(run.summary || {}).reduce((s, v) => s + (v.avgDurationMs || 0) * (v.successCount || 0), 0))}</td>
-      <td style="white-space:nowrap"><button onclick="deleteRun('${e(run.id)}',this)" class="btn-danger" data-i18n="deleteBtnText">${t('deleteBtnText', lang)}</button></td>
-    </tr>`;
+        const date = formatDateFromId(run.id, m.timestamp);
+        const variants = (m.variants || []).join(' · ');
+        const dur = Object.values(summary).reduce((s, v) => s + (v.avgDurationMs || 0) * (v.successCount || 0), 0);
+        return `<a class="run-card${verdictLvl ? ` verdict-${verdictLvl}` : ''}" href="/reports/${e(run.id)}${langQ}">
+      <div class="run-card-h">
+        ${statusPill}
+        <span class="run-card-id">${e(run.id)}</span>
+        ${date ? `<span class="run-card-date">${e(date)}</span>` : ''}
+      </div>
+      <div class="run-card-meta">${e(m.model || '-')} · ${m.sampleCount || 0} ${lang === 'zh' ? '用例' : 'samples'}${variants ? ` · ${e(variants)}` : ''}</div>
+      <div class="run-card-scores">${scoresHtml}</div>
+      <div class="run-card-foot">
+        <span class="run-card-stat">${fmtDuration(dur)}</span>
+        <button class="run-card-delete" onclick="event.preventDefault();event.stopPropagation();deleteRun('${e(run.id)}',this)" data-i18n="deleteBtnText" aria-label="${e(t('deleteBtnText', lang))}">×</button>
+      </div>
+    </a>`;
     }).join('');
     const runCount = lang === 'zh' ? `${runs.length} 次评测` : `${runs.length} runs`;
-    // 列表累计:只 sum 那些"全 variant 都报告了 cost"的 run,跳过任一 variant not reported 的。
-    // 排除 not reported(而不是把整体压成「—」)是因为列表常含数十个 run 仅 1-2 个是 codex,
-    // 把全部 sum 抹成「—」会丢掉绝大多数有效成本信息。
+    // v0.30 — 列表页移除累计金额展示(用户反馈无须暴露);保留 evaluationRuns 用于 trend links
     const evaluationRuns = runs.filter(isEvaluationReport);
-    const reportableRuns = evaluationRuns.filter((r) => Object.values(r.summary || {}).every((v) => v.execCostReported !== false));
-    const unmeasuredRunsCount = evaluationRuns.length - reportableRuns.length;
-    const totalCost = reportableRuns.reduce((s, r) => s + Object.values(r.summary || {}).reduce((sv, v) => sv + (v.totalExecCostUSD || 0), 0), 0);
-    // partial:有 reported 就显示数字 + tooltip;全部 not reported 才显示「—」;
-    // 全 reported(unmeasuredRunsCount=0)走老格式不包 span,保持 HTML snapshot 向后兼容。
-    // X/N 限定信息只放 tooltip,不挂在显眼位置 — 累计行视觉应该是单纯的数字。
-    const allUnmeasured = reportableRuns.length === 0 && runs.length > 0;
-    const baseLabel = lang === 'zh' ? '累计' : 'Total';
-    const costNumber = fmtCost(totalCost, !allUnmeasured);
-    let costLabel;
-    if (unmeasuredRunsCount === 0) {
-        // 全 reported → 跟旧版字节级一致,不破 snapshot
-        costLabel = `${baseLabel} ${costNumber}`;
-    }
-    else {
-        const tip = lang === 'zh'
-            ? `${reportableRuns.length}/${evaluationRuns.length} 次单次报告了 USD 成本;${unmeasuredRunsCount} 次 executor 不报(如 codex CLI),batch index 不计入累计以避免重复`
-            : `${reportableRuns.length}/${evaluationRuns.length} evaluation reports reported USD cost; ${unmeasuredRunsCount} excluded (executor doesn't report, e.g. codex CLI); batch indexes are excluded to avoid double counting`;
-        costLabel = `<span title="${e(tip)}">${baseLabel} ${costNumber}</span>`;
-    }
     // Collect variants with ≥2 reports for trend links
     const variantCounts = {};
     for (const run of evaluationRuns) {
@@ -244,36 +240,62 @@ export function renderRunList(runs, lang = DEFAULT_LANG) {
     return layout(t('title', lang), `
     <main>
     <h1>${t('title', lang)}</h1>
-    <p class="subtitle" data-i18n="subtitle">${t('subtitle', lang)} &middot; ${runCount} &middot; ${costLabel}</p>
+    <p class="subtitle" data-i18n="subtitle">${t('subtitle', lang)} &middot; ${runCount}</p>
     ${trendsSection}
     <div style="margin:12px 0">${skillHealthLink}</div>
     <div style="margin:12px 0;display:flex;gap:8px;align-items:center">
       <input id="filter-input" type="text" placeholder="${lang === 'zh' ? '搜索报告名称、变体...' : 'Filter by name, variant...'}" style="flex:1;max-width:320px;padding:6px 10px;font-size:13px;background:var(--bg-surface);color:var(--text-primary);border:1px solid var(--border);border-radius:var(--radius);outline:none" oninput="filterTable(this.value)">
       <span id="filter-count" style="font-size:11px;color:var(--text-muted)"></span>
     </div>
-    <div class="table-wrap">
-    <table id="report-table">
-      <thead><tr>
-        <th data-i18n="runId">${t('runId', lang)}</th>
-        <th data-i18n="model">${t('model', lang)}</th>
-        <th data-i18n="samples">${t('samples', lang)}</th>
-        <th data-i18n="score">${t('score', lang)}</th>
-        <th data-i18n="cost">${t('cost', lang)}</th>
-        <th>${lang === 'zh' ? '耗时' : 'Duration'}</th>
-        <th></th>
-      </tr></thead>
-      <tbody>${rows}</tbody>
-    </table>
-    </div>
+    ${renderScoringModal('guide-scoring-list', lang)}
+    <div id="run-list" class="run-list">${cards}</div>
+    <style>
+    /* 列表页 — 卡片式布局,顶部 verdict pill 主导,底部 hover 露出 delete */
+    .page-secondary-link { color:var(--text-secondary);font-size:13px;text-decoration:none;border:1px solid var(--border);padding:6px 14px;border-radius:var(--radius);display:inline-block;background:var(--bg-surface);transition:border-color .15s,color .15s }
+    .page-secondary-link:hover { color:var(--text-primary);border-color:var(--border-hover);text-decoration:none }
+    .run-list { display:flex;flex-direction:column;gap:12px;margin:16px 0 }
+    /* verdict 用左边粗带 + 浅色填充背景双重视觉提示,远看也能区分。
+       PROGRESS = sage 绿(进步),REGRESS = 砖红(回归),CAUTIOUS = 琥珀(警示),其他 = 灰(无信号) */
+    .run-card { display:block;padding:18px 22px;background:var(--bg-surface);border-radius:var(--radius);box-shadow:var(--shadow-sm);text-decoration:none;color:inherit;transition:box-shadow .15s,transform .05s;border-left:5px solid transparent;position:relative }
+    .run-card:hover { box-shadow:var(--shadow-md);text-decoration:none }
+    .run-card:hover .run-card-delete { opacity:1 }
+    .run-card.verdict-PROGRESS { border-left-color:var(--green);background:linear-gradient(90deg,var(--green-bg) 0%,var(--bg-surface) 80%) }
+    .run-card.verdict-REGRESS { border-left-color:var(--red);background:linear-gradient(90deg,var(--red-bg) 0%,var(--bg-surface) 80%) }
+    .run-card.verdict-CAUTIOUS { border-left-color:var(--yellow);background:linear-gradient(90deg,var(--yellow-bg) 0%,var(--bg-surface) 80%) }
+    .run-card.verdict-NOISE, .run-card.verdict-UNDERPOWERED { border-left-color:var(--text-faint);opacity:0.85 }
+    .run-card.verdict-SOLO { border-left-color:var(--accent);background:linear-gradient(90deg,var(--info-bg) 0%,var(--bg-surface) 80%) }
+    .run-card-h { display:flex;align-items:center;gap:12px;margin-bottom:6px;flex-wrap:wrap }
+    .run-card-id { font-size:15px;font-weight:600;color:var(--text-primary);font-family:"SF Mono",Menlo,Consolas,monospace;letter-spacing:-0.01em }
+    .run-card-date { font-size:12px;color:var(--text-muted);font-variant-numeric:tabular-nums;margin-left:auto }
+    .run-card-meta { font-size:13px;color:var(--text-secondary);margin:6px 0 12px;line-height:1.6 }
+    .run-card-scores { display:flex;flex-direction:column;gap:4px;margin:8px 0 }
+    .run-card-foot { display:flex;gap:16px;margin-top:10px;align-items:center;font-size:12px;color:var(--text-muted);font-variant-numeric:tabular-nums }
+    .run-card-stat { font-family:"SF Mono",Menlo,monospace }
+    .run-card-delete { margin-left:auto;background:transparent;border:none;color:var(--text-muted);font-size:18px;line-height:1;width:28px;height:28px;border-radius:14px;cursor:pointer;opacity:0;transition:opacity .15s,background .15s,color .15s;outline:none;appearance:none }
+    .run-card-delete:hover { background:var(--red-bg);color:var(--red) }
+    /* 分数行(每个 variant 一行) */
+    .rc-score-row { display:flex;align-items:center;gap:10px;font-size:13px;line-height:1.6 }
+    .rc-score-label { font-size:12px;color:var(--text-secondary);min-width:80px;text-align:right;overflow:hidden;text-overflow:ellipsis;white-space:nowrap;flex-shrink:0 }
+    .rc-score-bar { width:120px;height:6px;background:var(--bg-soft);border-radius:3px;flex-shrink:0;overflow:hidden }
+    .rc-score-bar-fill { height:100%;border-radius:3px;transition:width .2s }
+    .rc-score-bar-fill--pass { background:var(--green) }
+    .rc-score-bar-fill--warn { background:var(--yellow) }
+    .rc-score-bar-fill--fail { background:var(--red) }
+    .rc-score-num { font-size:13px;font-weight:600;font-variant-numeric:tabular-nums;font-family:"SF Mono",Menlo,monospace;min-width:36px }
+    .rc-score-num--pass { color:var(--green) }
+    .rc-score-num--warn { color:var(--yellow) }
+    .rc-score-num--fail { color:var(--red) }
+    .rc-score-empty { color:var(--text-faint);font-size:12px;font-style:italic }
+    </style>
     <script>
     function filterTable(q) {
-      var rows = document.querySelectorAll('#report-table tbody tr');
+      var cards = document.querySelectorAll('#run-list .run-card');
       var lower = q.toLowerCase();
       var shown = 0;
-      rows.forEach(function(row) {
-        var text = row.textContent.toLowerCase();
+      cards.forEach(function(card) {
+        var text = card.textContent.toLowerCase();
         var match = !q || text.indexOf(lower) !== -1;
-        row.style.display = match ? '' : 'none';
+        card.style.display = match ? '' : 'none';
         if (match) shown++;
       });
       var countEl = document.getElementById('filter-count');
@@ -285,7 +307,7 @@ export function renderRunList(runs, lang = DEFAULT_LANG) {
       fetch('/api/reports/' + encodeURIComponent(id), { method: 'DELETE' })
         .then(function(r) { return r.json(); })
         .then(function(d) {
-          if (d.ok) { btn.closest('tr').remove(); }
+          if (d.ok) { btn.closest('.run-card').remove(); }
           else { alert(I18N[lang].deleteFail + ': ' + (d.error || 'unknown')); }
         })
         .catch(function(err) { alert(I18N[lang].deleteFail + ': ' + err.message); });
@@ -309,9 +331,7 @@ export function renderRunDetail(report, lang = DEFAULT_LANG) {
     const summary = report.summary || {};
     const results = report.results || [];
     const cards = renderSummaryCards(variants, summary, lang, report.variance);
-    const pairwiseDiff = renderPairwiseDiff(report.meta.pairComparisons, lang);
-    const humanAgreement = renderHumanAgreement(report.meta.humanAgreement, lang);
-    const saturationCurve = renderSaturationCurve(report.variance?.saturation, variants, lang);
+    const methodologyAudit = renderMethodologyAudit(report, variants, summary, lang);
     const verdictPill = renderVerdictPill(report, lang);
     const sampleTable = renderSampleTable(variants, results, lang);
     const totalExecCost = Object.values(summary).reduce((s, v) => s + (v.totalExecCostUSD || 0), 0);
@@ -336,52 +356,41 @@ export function renderRunDetail(report, lang = DEFAULT_LANG) {
         zh: { baseline: '基线', skill: 'Skill', prompt: 'Prompt', agent: 'Agent', workflow: 'Workflow' },
         en: { baseline: 'Baseline', skill: 'Skill', prompt: 'Prompt', agent: 'Agent', workflow: 'Workflow' },
     };
+    // v0.30 — 实验配置弱化:单行 inline 摘要,跟 Agent 执行概览并排放(context-strip)。
+    // 每变体一行:左 3px 颜色条 + variant 名 + type/kind/source/strategy/cwd 用 · 分隔。
+    // 完整 key 名通过 hover tooltip 提供。视觉权重远低于综合分 hero。
     const variantConfigRows = (m.variantConfigs || []).map((config, i) => {
         const expTypeRaw = config.experimentType || '-';
         const expType = (typeLabels[lang] || typeLabels.en)[String(expTypeRaw)] || expTypeRaw;
         const source = config.artifactKind === 'baseline'
-            ? (lang === 'zh' ? '无' : 'None')
+            ? (lang === 'zh' ? '无来源' : 'no source')
             : (sourceLabels[lang] || sourceLabels.en)[config.artifactSource] || config.artifactSource;
         const strategy = (strategyLabels[lang] || strategyLabels.en)[config.executionStrategy] || config.executionStrategy;
         const cwdRaw = config.cwd || '';
         const runtimeContext = cwdRaw
             ? cwdRaw.replace(/.*\/Projects\//, '').replace(/.*\/Documents\//, '').replace(/\/Users\/[^/]+\//, '~/')
-            : (lang === 'zh' ? '默认' : 'default');
+            : (lang === 'zh' ? '默认 cwd' : 'default cwd');
         const color = COLORS[i % COLORS.length];
-        return `<tr>
-      <td style="border-left:3px solid ${color};padding-left:12px"><strong>${e(config.variant)}</strong></td>
-      <td>${e(expType)}</td>
-      <td>${e((artifactKindLabels[lang] || artifactKindLabels.en)[config.artifactKind] || config.artifactKind)}</td>
-      <td>${e(source)}</td>
-      <td>${e(strategy)}</td>
-      <td title="${e(cwdRaw)}">${e(runtimeContext)}</td>
-    </tr>`;
+        const artifactKindLabel = e((artifactKindLabels[lang] || artifactKindLabels.en)[config.artifactKind] || config.artifactKind);
+        const tooltip = `${t('variantType', lang)}: ${expType} · ${t('variantArtifactKind', lang)}: ${artifactKindLabel} · ${t('variantArtifactSource', lang)}: ${source} · ${t('variantExecutionStrategy', lang)}: ${strategy} · ${t('variantRuntimeContext', lang)}: ${cwdRaw || (lang === 'zh' ? '默认' : 'default')}`;
+        return `<div class="ctx-row" style="border-left:3px solid ${color}" title="${e(tooltip)}">
+      <span class="ctx-row-name">${e(config.variant)}</span>
+      <span class="ctx-row-bits">${e(expType)} <span class="ctx-sep">·</span> ${artifactKindLabel} <span class="ctx-sep">·</span> ${e(source)} <span class="ctx-sep">·</span> ${e(strategy)} <span class="ctx-sep">·</span> ${e(runtimeContext)}</span>
+    </div>`;
     }).join('');
     const configModalId = 'guide-variant-config';
     const repeatSuffix = report.variance
         ? (lang === 'zh' ? ` × ${report.variance.runs} 轮` : ` × ${report.variance.runs} runs`)
         : '';
     const experimentSummary = lang === 'zh'
-        ? `${m.sampleCount} 个评测用例 × ${variants.length} 组实验${repeatSuffix}`
+        ? `${m.sampleCount} 个用例 × ${variants.length} 组实验${repeatSuffix}`
         : `${pluralizeEn(m.sampleCount, 'sample')} × ${pluralizeEn(variants.length, 'variant')}${repeatSuffix}`;
     const variantConfigSection = variantConfigRows ? `
-    <section style="margin:20px 0">
-      <h2 style="display:flex;align-items:center;gap:4px">${t('variantConfig', lang)} <button type="button" class="hint-btn" onclick="openModal('${configModalId}')" aria-label="${e(t('variantConfigDesc', lang))}" aria-haspopup="dialog">?</button></h2>
-      <p style="font-size:12px;color:var(--text-muted);margin-bottom:8px">${experimentSummary}</p>
-      <div class="table-wrap">
-        <table>
-          <thead><tr>
-            <th>🏷️ ${t('variants', lang)}</th>
-            <th>🧪 ${t('variantType', lang)}</th>
-            <th>📦 ${t('variantArtifactKind', lang)}</th>
-            <th>📂 ${t('variantArtifactSource', lang)}</th>
-            <th>⚙️ ${t('variantExecutionStrategy', lang)}</th>
-            <th>🖥️ ${t('variantRuntimeContext', lang)}</th>
-          </tr></thead>
-          <tbody>${variantConfigRows}</tbody>
-        </table>
-      </div>
-    </section>
+    <div class="ctx-block">
+      <div class="ctx-h">${t('variantConfig', lang)} <button type="button" class="hint-btn" onclick="openModal('${configModalId}')" aria-label="${e(t('variantConfigDesc', lang))}" aria-haspopup="dialog">?</button></div>
+      <div class="ctx-sub">${experimentSummary}</div>
+      <div class="ctx-rows">${variantConfigRows}</div>
+    </div>
     <div id="${configModalId}" class="modal-overlay" role="dialog" aria-modal="true" aria-labelledby="${configModalId}-title" onclick="if(event.target===this)closeModal('${configModalId}')">
       <div class="modal-content">
         <div class="modal-header">
@@ -409,14 +418,20 @@ export function renderRunDetail(report, lang = DEFAULT_LANG) {
       </div>
     </div>
   ` : '';
-    return layout(`${report.id}`, `
-    <main>
-    <nav class="nav"><a href="/${langQ}" data-i18n="backToList">${t('backToList', lang)}</a></nav>
-    <h1>${e(report.id)}</h1>
-    ${verdictPill}
-    <div class="meta-tags">
-      <span class="meta-tag">${t('model', lang)}: ${e(m.model)}</span>
-      ${(() => {
+    // v0.30 重设计 — 评分/单测各自独立 title + meta strip(不共享头部)。
+    // 顶部只保留极简身份带:返回链接 + run-id pill。完整 H1 + meta tags 移进各自 tab 面板。
+    const idStamp = (() => {
+        const idMatch = /^.+?-(\d{8})-(\d{4})$/.exec(report.id);
+        if (idMatch) {
+            const [, ymd, hm] = idMatch;
+            return `${ymd.slice(0, 4)}-${ymd.slice(4, 6)}-${ymd.slice(6, 8)} ${hm.slice(0, 2)}:${hm.slice(2, 4)}`;
+        }
+        return '';
+    })();
+    // ──────────── 评分 tab 的 meta tags(原版全部留这里) ────────────
+    const scoreMetaTags = `<div class="meta-tags">
+    <span class="meta-tag">${t('model', lang)}: ${e(m.model)}</span>
+    ${(() => {
         if (m.noJudge)
             return `<span class="meta-tag">${t('judge', lang)}: none</span>`;
         const list = m.judgeModels ?? [];
@@ -426,39 +441,125 @@ export function renderRunDetail(report, lang = DEFAULT_LANG) {
             return `<span class="meta-tag">${t('judge', lang)}: ${e(`${list[0].executor}:${list[0].model}`)}</span>`;
         return `<span class="meta-tag" title="${t('ensembleDesc', lang)}">${t('judgeModelsLabel', lang)}: ${list.map((j) => e(`${j.executor}:${j.model}`)).join(' · ')}</span>`;
     })()}
-      ${m.judgeRepeat && m.judgeRepeat > 1 ? `<span class="meta-tag" title="${t('judgeStddevDesc', lang)}">${t('judgeRepeatLabel', lang)}: ${m.judgeRepeat}</span>` : ''}
-      <span class="meta-tag">${t('executor', lang)}: ${e(m.executor || 'claude')}</span>
-      <span class="meta-tag"${execCostReported ? '' : ` title="${e(lang === 'zh' ? 'executor 不报 USD 成本(如 codex CLI),无法估算' : 'executor does not report USD cost (e.g. codex CLI); not measurable')}"`}>${t('cost', lang)}: ${fmtCost(totalExecCost, execCostReported)}</span>
-      <span class="meta-tag"${totalCostReported ? '' : ` title="${e(costCompletenessTooltip(lang))}"`}>${t('totalCost', lang)}: ${fmtKnownCost(m.totalCostUSD, totalCostReported)}</span>
-      <span class="meta-tag">${lang === 'zh' ? '耗时' : 'Duration'}: ${fmtDuration(totalDurationMs)}</span>
-      ${m.gitInfo ? `<span class="meta-tag">${lang === 'zh' ? '提交' : 'commit'}: ${e(m.gitInfo.commitShort)}${m.gitInfo.dirty ? '*' : ''} (${e(m.gitInfo.branch)})</span>` : ''}
-      ${m.judgePromptHash ? `<span class="meta-tag" title="${t('judgePromptHashDesc', lang)}">${t('judgePromptHashLabel', lang)}: <code>${e(m.judgePromptHash)}</code></span>` : ''}${renderExecutorRuntimeTags(m, lang)}${renderJudgeRuntimeTags(m, lang)}
-      ${m.sampleHashes ? `<span class="meta-tag" style="color:var(--text-muted)" title="${t('sampleHashCountDesc', lang)}">${t('sampleHashCount', lang)}: ${Object.keys(m.sampleHashes).length}/${m.sampleCount}</span>` : ''}
-      ${m.evaluationFramework ? `<span class="meta-tag" title="${t('evalFrameworkDesc', lang)}">${t('evalFrameworkLabel', lang)}: ${m.evaluationFramework === 'bootstrap' ? t('evalFrameworkBootstrap', lang) : m.evaluationFramework === 'both' ? t('evalFrameworkBoth', lang) : t('evalFrameworkTTest', lang)}</span>` : ''}
-      ${renderDebiasModeTag(m.debiasMode, lang)}
-      ${m.blind ? `<span class="meta-tag" style="color:var(--green)" data-i18n="blindLabel">${t('blindLabel', lang)}</span>` : ''}
-    </div>
-    ${m.blind ? `
+    ${m.judgeRepeat && m.judgeRepeat > 1 ? `<span class="meta-tag" title="${t('judgeStddevDesc', lang)}">${t('judgeRepeatLabel', lang)}: ${m.judgeRepeat}</span>` : ''}
+    <span class="meta-tag">${t('executor', lang)}: ${e(m.executor || 'claude')}</span>
+    ${m.effort ? `<span class="meta-tag" title="${e(lang === 'zh' ? 'executor LLM 的扩展思考预算(--effort)。跨 effort 报告不可严格比较' : 'reasoning effort for executor LLM (--effort); reports across different efforts are not strictly comparable')}">effort: ${e(m.effort)}</span>` : ''}
+    <span class="meta-tag"${execCostReported ? '' : ` title="${e(lang === 'zh' ? 'executor 不报 USD 成本(如 codex CLI),无法估算' : 'executor does not report USD cost (e.g. codex CLI); not measurable')}"`}>${t('cost', lang)}: ${fmtCost(totalExecCost, execCostReported)}</span>
+    <span class="meta-tag"${totalCostReported ? '' : ` title="${e(costCompletenessTooltip(lang))}"`}>${t('totalCost', lang)}: ${fmtKnownCost(m.totalCostUSD, totalCostReported)}</span>
+    <span class="meta-tag">${lang === 'zh' ? '耗时' : 'Duration'}: ${fmtDuration(totalDurationMs)}</span>
+    ${m.gitInfo ? `<span class="meta-tag">${lang === 'zh' ? '提交' : 'commit'}: ${e(m.gitInfo.commitShort)}${m.gitInfo.dirty ? '*' : ''} (${e(m.gitInfo.branch)})</span>` : ''}
+    ${m.sampleHashes ? `<span class="meta-tag" title="${t('sampleHashCountDesc', lang)}">${t('sampleHashCount', lang)}: ${Object.keys(m.sampleHashes).length}/${m.sampleCount}</span>` : ''}
+    ${m.evaluationFramework ? `<span class="meta-tag" title="${t('evalFrameworkDesc', lang)}">${t('evalFrameworkLabel', lang)}: ${m.evaluationFramework === 'bootstrap' ? t('evalFrameworkBootstrap', lang) : m.evaluationFramework === 'both' ? t('evalFrameworkBoth', lang) : t('evalFrameworkTTest', lang)}</span>` : ''}
+    ${renderDebiasModeTag(m.debiasMode, lang)}
+    ${m.blind ? `<span class="meta-tag" style="color:var(--green)" data-i18n="blindLabel">${t('blindLabel', lang)}</span>` : ''}
+  </div>`;
+    const auditFingerprints = (() => {
+        const auditTags = [
+            m.judgePromptHash ? `<span class="meta-tag" title="${t('judgePromptHashDesc', lang)}">${t('judgePromptHashLabel', lang)}: <code>${e(m.judgePromptHash)}</code></span>` : '',
+            renderRuntimeFingerprintTags(m, lang),
+        ].join('');
+        if (!auditTags)
+            return '';
+        return `<details class="audit-fingerprints"><summary>${lang === 'zh' ? '审计指纹（用于复现校验）' : 'Audit fingerprints (for reproducibility)'}</summary><div class="meta-tags">${auditTags}</div></details>`;
+    })();
+    const blindReveal = m.blind ? `
     <div style="margin:12px 0">
       <button onclick="document.getElementById('blind-reveal').style.display=document.getElementById('blind-reveal').style.display==='none'?'block':'none'" data-i18n="revealBlind">${t('revealBlind', lang)}</button>
       <div id="blind-reveal" style="display:none;margin-top:8px;padding:12px;background:var(--bg-surface);border:1px solid var(--border);border-radius:var(--radius)" role="region" aria-label="Blind variant mapping">
         ${Object.entries(m.blindMap || {}).map(([label, real]) => `<div style="font-size:13px;color:var(--text-secondary)"><strong>Variant ${e(label)}</strong> → ${e(real)}</div>`).join('')}
       </div>
-    </div>` : ''}
+    </div>` : '';
+    // ──────────── 单测 tab 的 meta tags(只有 sample 维度数据,不含 cost / model) ────────────
+    const sampleSnaps = report.sampleSnapshots ?? {};
+    let testPassed = 0;
+    let testFailed = 0;
+    let testTripwire = 0;
+    for (const r of (report.results || [])) {
+        const v = m.variants?.[0] ? r.variants[m.variants[0]] : null;
+        if (!v)
+            continue;
+        const passed = (v.assertions?.details ?? []).every((d) => d.passed);
+        if (passed)
+            testPassed += 1;
+        else if (sampleSnaps[r.sample_id]?.tripwire || (v.diagnostic?.rootCause ?? []).includes('tripwire_intentional'))
+            testTripwire += 1;
+        else
+            testFailed += 1;
+    }
+    const testMetaTags = `<div class="meta-tags">
+    <span class="meta-tag">${m.sampleCount} ${lang === 'zh' ? '用例' : 'samples'}</span>
+    <span class="meta-tag" style="color:var(--green)">✓ ${testPassed} ${lang === 'zh' ? '通过' : 'passed'}</span>
+    ${testFailed > 0 ? `<span class="meta-tag" style="color:var(--red)">✗ ${testFailed} ${lang === 'zh' ? '失败' : 'failed'}</span>` : ''}
+    ${testTripwire > 0 ? `<span class="meta-tag" style="color:#7a6b89">◆ ${testTripwire} ${lang === 'zh' ? '诱错触发' : 'tripwire'}</span>` : ''}
+    <span class="meta-tag">${(m.variants || []).join(' · ')}</span>
+  </div>`;
+    return layout(`${report.id}`, `
+    <main>
+    <nav class="nav"><a href="/${langQ}" data-i18n="backToList">${t('backToList', lang)}</a></nav>
+    <div class="report-id-line">
+      <code class="report-id-pill">${e(report.id)}</code>
+      ${idStamp ? `<span class="report-id-stamp">${e(idStamp)}</span>` : ''}
+    </div>
 
-    ${variantConfigSection}
+    <div class="omk-view-tabs" role="tablist">
+      <button type="button" class="omk-view-tab omk-view-tab--active" data-view="score" onclick="omkSwitchView('score')">${lang === 'zh' ? '📊 评测视角' : '📊 Score view'}</button>
+      <button type="button" class="omk-view-tab" data-view="test" onclick="omkSwitchView('test')">${lang === 'zh' ? '✅ 功能视角' : '✅ Functional view'}</button>
+    </div>
+
+    <div class="omk-view-panel" data-view="score">
+    <p class="subtitle">${lang === 'zh' ? '主观质量打分(judge)+ 客观断言聚合,产出综合评分与 verdict' : 'Subjective judge + objective assertions aggregated into composite + verdict'}</p>
+    ${verdictPill}
+    ${scoreMetaTags}
+    ${blindReveal}
 
-    <section>${cards}${pairwiseDiff}${humanAgreement}${saturationCurve}</section>
+    <section>${cards}</section>
+    <div class="ctx-strip">${variantConfigSection}${renderAgentOverview(variants, summary, lang)}</div>
+    ${auditFingerprints}
+    ${methodologyAudit}
 
     ${renderVarianceComparisons(report.variance, lang, Boolean(report.meta.layeredStats), summary)}
 
     ${renderAnalysis(report, lang)}
 
-    ${renderAgentOverview(variants, summary, lang)}
-
     ${renderKnowledgeInteractionSection(report.analysis?.coverage, report.analysis?.gapReports, lang)}
 
     <section>${sampleTable}</section>
+    </div>
+
+    <div class="omk-view-panel" data-view="test" hidden>
+    <p class="subtitle">${lang === 'zh' ? '每条用例当一条功能测试看:期望(rubric/assertions/mocks)、实际执行(trace)、失败时给诊断建议' : 'Each sample as a functional test: expected(rubric/assertions/mocks), actual(trace), diagnostic suggestions on failure'}</p>
+    ${testMetaTags}
+    ${renderTestView(report, lang)}
+    </div>
+
+    <style>${TEST_VIEW_CSS}
+/* Context strip — 实验配置 + Agent 概览并排,作为辅助元数据,视觉权重低于综合分 hero / 六维 / 用例详情 */
+.ctx-strip { display:grid;gap:14px;grid-template-columns:1fr 1fr;margin:14px 0 24px }
+@media (max-width:760px) { .ctx-strip { grid-template-columns:1fr } }
+.ctx-block { background:var(--bg-surface);border-radius:var(--radius);box-shadow:var(--shadow-sm);padding:12px 16px }
+.ctx-h { font-size:13px;color:var(--text-secondary);font-weight:600;margin-bottom:2px;display:flex;align-items:center;gap:4px;letter-spacing:0.02em }
+.ctx-sub { font-size:12px;color:var(--text-muted);margin-bottom:10px }
+.ctx-rows { display:flex;flex-direction:column;gap:6px }
+.ctx-row { padding:8px 12px;background:var(--bg-soft);border-radius:4px;display:flex;align-items:center;gap:10px;font-size:13px;line-height:1.5;flex-wrap:wrap }
+.ctx-row-name { font-weight:600;color:var(--text-primary);font-family:"SF Mono",Menlo,monospace;font-size:12.5px;letter-spacing:-0.01em }
+.ctx-row-bits { color:var(--text-secondary);flex:1;min-width:0;font-size:12.5px }
+.ctx-sep { color:var(--text-muted);margin:0 2px;opacity:0.6 }
+/* 报告身份带 — 极简一行,run-id pill + 时间戳 + verdict pill。完整 H1 + meta 移进各 tab 自己的 panel。 */
+.report-id-line { display:flex;align-items:center;gap:12px;flex-wrap:wrap;margin:8px 0 18px }
+.report-id-pill { font-size:14px;font-family:"SF Mono",Menlo,monospace;color:var(--text-primary);background:var(--bg-surface);padding:5px 12px;border-radius:14px;border:1px solid var(--border);letter-spacing:-0.01em;font-weight:500 }
+.report-id-stamp { font-size:12px;color:var(--text-muted);font-variant-numeric:tabular-nums }
+/* 顶部 tab 切换器 — 显式 reset border-radius / appearance / outline,
+   防止全局 button 规则给底部加奇怪圆角和深色背景。 */
+.omk-view-tabs { display:flex;gap:0;margin:16px 0 24px;border-bottom:1px solid var(--border) }
+.omk-view-tab { padding:12px 20px;background:transparent;border:none;border-radius:0;cursor:pointer;color:var(--text-secondary);border-bottom:2px solid transparent;font-size:14px;font-weight:500;outline:none;appearance:none;-webkit-appearance:none;margin-bottom:-1px;font-family:inherit }
+.omk-view-tab--active { color:var(--text-primary);border-bottom-color:var(--accent);font-weight:600 }
+.omk-view-tab:hover:not(.omk-view-tab--active) { color:var(--text-primary) }
+.omk-view-tab:focus-visible { outline:2px solid var(--accent);outline-offset:-4px;border-radius:4px }
+/* tab 面板内的 H1 — 跟列表 H1 同级别但加 emoji 区分视角 */
+.omk-view-panel h1 { font-size:1.5rem;margin:0 0 6px }
+.omk-view-panel .subtitle { font-size:14px;color:var(--text-secondary);margin:0 0 20px;line-height:1.6 }
+    </style>
+    <script>${TEST_VIEW_JS}</script>
 
     </main>
   `, lang);
@@ -517,9 +618,14 @@ export function renderBatchEvaluationDetail(report, lang = DEFAULT_LANG) {
         return `<span class="meta-tag" title="${t('ensembleDesc', lang)}">${t('judgeModelsLabel', lang)}: ${list.map((j) => e(`${j.executor}:${j.model}`)).join(' · ')}</span>`;
     })()}
       <span class="meta-tag">${t('executor', lang)}: ${e(m.executor || 'claude')}</span>
-      ${renderExecutorRuntimeTags(m, lang)}${renderJudgeRuntimeTags(m, lang)}
       <span class="meta-tag"${totalCostReported ? '' : ` title="${e(costCompletenessTooltip(lang))}"`}>${t('totalCost', lang)}: ${fmtKnownCost(m.totalCostUSD, totalCostReported)}</span>
     </div>
+    ${(() => {
+        const auditTags = renderRuntimeFingerprintTags(m, lang);
+        if (!auditTags)
+            return '';
+        return `<details class="audit-fingerprints"><summary>${lang === 'zh' ? '审计指纹（用于复现校验）' : 'Audit fingerprints (for reproducibility)'}</summary><div class="meta-tags">${auditTags}</div></details>`;
+    })()}
 
     <section>
     <h2>${t('batchOverview', lang)}</h2>