dsh-llm-verifier 0.1.17 → 0.2.0

package/lib/client.js

@@ -32,9 +32,9 @@ window.__ModuleLoader__.load({
 			"field.autoRouteMaxCandidates.title": "最多候选数",
 			"field.autoRouteMaxCandidates.help": "一次自动 select 最多纳入的真实候选数量，至少 3。",
 			"field.autoRouteMaxPerTask.title": "每任务最多路由",
-			"field.autoRouteMaxPerTask.help": "compare/select/track 自动路由的任务预算；指纹去重后仍受此上限约束。",
+			"field.autoRouteMaxPerTask.help": "自动路由（compare/select/track 与语义分类）的任务尝试上限；与「每任务最多验收」共享同一个计数器，实际可用次数是两者之和。",
 			"field.autoRouteMaxPerSession.title": "每会话最多路由",
-			"field.autoRouteMaxPerSession.help": "同一会话中 compare/select/track 自动路由总预算。",
+			"field.autoRouteMaxPerSession.help": "同一会话中自动路由的尝试上限；与「每会话最多验收」共享计数器，实际可用次数同样是两者之和。",
 			"field.autoTrackCompletionThreshold.title": "进度完成阈值",
 			"field.autoTrackCompletionThreshold.help": "track 任一检查点低于该值时 steering 要求 Agent 继续执行；范围 0–1。",
 			"field.autoRouteMaxItemChars.title": "单项证据字符上限",
@@ -47,17 +47,21 @@ window.__ModuleLoader__.load({
 			"field.autoMaxModelCallsPerSession.help": "整个会话所有自动验证阶段共享的估算模型调用上限。",
 			"field.autoVerifyThreshold.title": "通过阈值",
 			"field.autoVerifyThreshold.help": "会话证据分数达到该值且胜过空工作基线才允许结束；范围 0–1。",
-			"field.autoVerifyRepeats.title": "自动评估轮次",
-			"field.autoVerifyRepeats.help": "每项标准的自动评分重复次数；1 为低成本初筛。",
+			"field.autoVerifyRepeats.title": "路由评估轮次",
+			"field.autoVerifyRepeats.help": "自动路由（compare / select / track）每项标准的评分重复次数；1 为低成本初筛。",
+			"field.autoVerifyFinalRepeats.title": "最终验收轮次",
+			"field.autoVerifyFinalRepeats.help": "最终会话验收每项标准的评分重复次数；偶数轮会交换 A/B 位置以抵消位置偏好，默认 2。",
 			"field.autoVerifyMinToolCalls.title": "智能模式最少工具调用",
-			"field.autoVerifyMinToolCalls.help": "达到此工具证据数量且包含写入/执行类操作时才自动验收。",
+			"field.autoVerifyMinToolCalls.help": "统计除 Verifier 之外的全部工具调用（含 read/grep 等只读工具）；此外还必须存在写入/执行类操作与至少一条成功结果。",
 			"field.autoVerifyMaxChars.title": "最大证据字符",
 			"field.autoVerifyMaxChars.help": "发送给裁判前保留的最近会话证据字符数。",
 			"field.autoVerifyMaxPerTask.title": "每任务最多验收",
-			"field.autoVerifyMaxPerTask.help": "低分反馈后允许再次验收的次数上限，防止循环。",
+			"field.autoVerifyMaxPerTask.help": "最终验收的任务尝试上限；与「每任务最多路由」共享同一个计数器，实际可用次数是两者之和，用于防止低分反馈形成无限修复循环。",
 			"field.autoVerifyMaxPerSession.title": "每会话最多验收",
-			"field.autoVerifyMaxPerSession.help": "同一会话中的自动验收总预算。",
+			"field.autoVerifyMaxPerSession.help": "同一会话中最终验收的尝试上限；与「每会话最多路由」共享计数器，实际可用次数同样是两者之和。",
 			"field.autoVerify.warnNotice": "自动路由按 select → compare → track → current_session 的阶段顺序处理；脱敏后的任务、Assistant 轨迹与真实工具输出会发送给所选裁判模型。候选选择或进度不足会 steering 继续工作，最终验收未通过会要求修复并重新验证。",
+			"field.autoMaxModelCallsPerTask.warnBudget": "当前单任务预算为 {current} 次模型调用，低于 {judges} 位裁判完整锦标赛所需的最差情况（{required} 次）。建议调大该预算以避免自动路由过早耗尽。",
+			"field.autoMaxModelCallsPerSession.warnBudget": "当前会话预算为 {current} 次模型调用，低于 {judges} 位裁判推荐的最差情况（{required} 次）。建议调大该预算以避免会话验证过早耗尽。",
 			"section.model": "裁判模型",
 			"field.provider.title": "供应商",
 			"field.provider.help": "当前 DSH 中可路由的模型供应商。",
@@ -68,13 +72,35 @@ window.__ModuleLoader__.load({
 			"field.reasoningEffort.default": "模型默认",
 			"field.maxTokens.title": "最大输出 Token",
 			"field.maxTokens.help": "单次裁判请求允许生成的最大 Token 数。",
+			"field.temperature.title": "裁判采样温度",
+			"field.temperature.help": "裁判模型调用的采样温度，范围 0–2；较低值可保证判决的一致与可复现性，默认值为 0.2。",
+			"field.label.title": "裁判标签",
+			"field.label.help": "在统计看板和明细中显示的裁判名称，留空时默认使用模型名称。",
+			"field.label.placeholder": "默认使用模型名称",
+			"field.extraJudges.title": "附加裁判",
+			"field.extraJudges.help": "添加最多 4 个额外裁判模型组成评审团。同一供应商与模型不能重复，且不能与主裁判相同。",
+			"field.extraJudges.add": "添加裁判",
+			"field.extraJudges.remove": "删除",
+			"field.extraJudges.labelTitle": "标签",
+			"field.extraJudges.labelPlaceholder": "自定义标签（可选）",
+			"field.extraJudges.providerAria": "附加裁判 #{index} 供应商",
+			"field.extraJudges.modelAria": "附加裁判 #{index} 模型",
+			"field.extraJudges.effortAria": "附加裁判 #{index} 推理强度",
+			"field.extraJudges.labelAria": "附加裁判 #{index} 标签",
+			"field.extraJudges.removeAria": "删除附加裁判 #{index}",
+			"field.extraJudges.conflictPrimary": "附加裁判 #{index} 与主裁判模型重复（{id}），请选择不同模型。",
+			"field.extraJudges.conflictDuplicate": "附加裁判 #{index} 与附加裁判 #{other} 重复（{id}），请选择不同模型。",
 			"section.execution": "执行",
 			"field.maxConcurrency.title": "最大并发",
 			"field.maxConcurrency.help": "所有 verifier 工具共享的请求并发上限。",
 			"field.maxRetries.title": "最多重试",
 			"field.maxRetries.help": "限流、超时和短暂网络错误的重试次数。",
+			"field.retryBaseDelayMs.title": "重试基础延迟",
+			"field.retryBaseDelayMs.help": "遇到限流或暂时网络错误时重试的基础等待时间，单位为毫秒。",
 			"field.timeoutMs.title": "请求超时",
 			"field.timeoutMs.help": "单次模型请求的超时时间，单位为毫秒。",
+			"field.cacheDir.title": "缓存相对目录",
+			"field.cacheDir.help": "评分持久化缓存存放的相对目录路径（相对于话题目录），不能为空。",
 			"field.cacheMaxEntries.title": "缓存条目上限",
 			"field.cacheMaxEntries.help": "本地持久评分缓存保留的最大条目数。",
 			"section.cost": "费用估算",
@@ -101,6 +127,24 @@ window.__ModuleLoader__.load({
 			"stats.successCalls": "成功调用",
 			"stats.failedCalls": "失败调用",
 			"stats.avgDuration": "平均耗时",
+			"stats.phase.explicit": "显式调用",
+			"stats.phase.compare": "两项对比",
+			"stats.phase.select": "多项优选",
+			"stats.phase.track": "进度跟踪",
+			"stats.phase.final": "最终验收",
+			"stats.phase.plan_review": "计划预审",
+			"stats.phase.team_task": "Team 任务",
+			"stats.phase.semantic": "语义路由",
+			"stats.outcome.passed": "通过",
+			"stats.outcome.below-threshold": "未达标",
+			"stats.outcome.tie": "平局",
+			"stats.outcome.error": "异常",
+			"stats.outcome.dropped-over-budget": "超预算丢弃",
+			"stats.outcome.invalid-references": "引用无效",
+			"stats.winner.tie": "平局",
+			"stats.verdict.winner": "胜方 {winner}",
+			"stats.verdict.scoreThreshold": "得分 {score}（阈值 {threshold}）",
+			"stats.verdict.scoreOnly": "得分 {score}",
 			"metric.cacheHitRate": "缓存命中率",
 			"metric.cacheHitNote": "{hits} 命中 / {total} 评分",
 			"metric.tokens": "Token",
@@ -168,9 +212,9 @@ window.__ModuleLoader__.load({
 			"field.autoRouteMaxCandidates.title": "Max Candidates",
 			"field.autoRouteMaxCandidates.help": "Maximum number of real candidates included in an automatic select, at least 3.",
 			"field.autoRouteMaxPerTask.title": "Max Routes per Task",
-			"field.autoRouteMaxPerTask.help": "Task budget for automatic compare/select/track routing; still constrained by this cap after fingerprint deduplication.",
+			"field.autoRouteMaxPerTask.help": "Task attempt cap for automatic routing (compare/select/track and semantic classification); shares one counter with \"Max Verifications per Task\", so the effective limit is the sum of both.",
 			"field.autoRouteMaxPerSession.title": "Max Routes per Session",
-			"field.autoRouteMaxPerSession.help": "Total budget for automatic compare/select/track routing within the same session.",
+			"field.autoRouteMaxPerSession.help": "Session attempt cap for automatic routing; shares its counter with \"Max Verifications per Session\" (effective limit is the sum of both).",
 			"field.autoTrackCompletionThreshold.title": "Progress Completion Threshold",
 			"field.autoTrackCompletionThreshold.help": "When any track checkpoint falls below this score, steering instructs the Agent to continue; range 0–1.",
 			"field.autoRouteMaxItemChars.title": "Max Chars per Route Item",
@@ -181,19 +225,23 @@ window.__ModuleLoader__.load({
 			"field.autoMaxModelCallsPerTask.help": "Estimated model call cap shared across semantic classification, intermediate routing, and final acceptance per task.",
 			"field.autoMaxModelCallsPerSession.title": "Model Call Budget per Session",
 			"field.autoMaxModelCallsPerSession.help": "Estimated model call cap shared across all automatic verification phases throughout the entire session.",
+			"field.autoVerifyRepeats.title": "Routing Repeats",
+			"field.autoVerifyRepeats.help": "Scoring repeats per criterion for automatic routing (compare / select / track); 1 is low-cost preliminary screening.",
+			"field.autoVerifyFinalRepeats.title": "Final Acceptance Repeats",
+			"field.autoVerifyFinalRepeats.help": "Scoring repeats per criterion for the final session acceptance; even rounds swap A/B positions to cancel position bias. Defaults to 2.",
 			"field.autoVerifyThreshold.title": "Pass Threshold",
 			"field.autoVerifyThreshold.help": "Evidence score must meet this threshold and beat the empty-work baseline to allow completion; range 0–1.",
-			"field.autoVerifyRepeats.title": "Auto Evaluation Repeats",
-			"field.autoVerifyRepeats.help": "Number of scoring repeats per criterion for auto evaluation; 1 is low-cost preliminary screening.",
 			"field.autoVerifyMinToolCalls.title": "Min Tool Calls (Smart Mode)",
-			"field.autoVerifyMinToolCalls.help": "Auto-verification triggers only when reaching this tool evidence count and containing write/exec operations.",
+			"field.autoVerifyMinToolCalls.help": "Counts every non-Verifier tool call, read-only tools included; consequential write/exec work and at least one successful result are still required.",
 			"field.autoVerifyMaxChars.title": "Max Evidence Characters",
 			"field.autoVerifyMaxChars.help": "Maximum recent session evidence characters retained before sending to the judge.",
 			"field.autoVerifyMaxPerTask.title": "Max Verifications per Task",
-			"field.autoVerifyMaxPerTask.help": "Maximum verification attempts allowed after low-score feedback to prevent infinite loops.",
+			"field.autoVerifyMaxPerTask.help": "Final-acceptance attempt cap per task; shares one counter with \"Max Routes per Task\", so the effective limit is the sum of both. Prevents endless low-score repair loops.",
 			"field.autoVerifyMaxPerSession.title": "Max Verifications per Session",
-			"field.autoVerifyMaxPerSession.help": "Total auto-verification budget within the same session.",
+			"field.autoVerifyMaxPerSession.help": "Final-acceptance attempt cap within one session; shares its counter with \"Max Routes per Session\" (effective limit is the sum of both).",
 			"field.autoVerify.warnNotice": "Automatic routing follows the phase order: select → compare → track → current_session. Sanitized tasks, Assistant trajectories, and real tool outputs are sent to the judge model. Candidate selection or insufficient progress steers the Agent to keep working; failed final acceptance requires remediation and re-verification.",
+			"field.autoMaxModelCallsPerTask.warnBudget": "Current task budget ({current} calls) is below the worst-case requirement ({required} calls) for {judges} judge(s) in a full tournament. Consider increasing it to avoid premature budget exhaustion.",
+			"field.autoMaxModelCallsPerSession.warnBudget": "Current session budget ({current} calls) is below the worst-case requirement ({required} calls) for {judges} judge(s). Consider increasing it to avoid premature budget exhaustion.",
 			"section.model": "Judge Model",
 			"field.provider.title": "Provider",
 			"field.provider.help": "Routable model providers currently available in DSH.",
@@ -204,13 +252,35 @@ window.__ModuleLoader__.load({
 			"field.reasoningEffort.default": "Model Default",
 			"field.maxTokens.title": "Max Output Tokens",
 			"field.maxTokens.help": "Maximum tokens allowed in a single judge model completion.",
+			"field.temperature.title": "Judge Sampling Temperature",
+			"field.temperature.help": "Sampling temperature for judge model calls, range 0–2. Lower values make judge verdicts reproducible. Default is 0.2.",
+			"field.label.title": "Judge Label",
+			"field.label.help": "Display name for the judge in statistics and dashboards. Defaults to the model name when empty.",
+			"field.label.placeholder": "Defaults to model name",
+			"field.extraJudges.title": "Extra Judges",
+			"field.extraJudges.help": "Add up to 4 extra judge models for ensemble verification. Duplicate provider and model combinations are not allowed, including the primary judge.",
+			"field.extraJudges.add": "Add Judge",
+			"field.extraJudges.remove": "Remove",
+			"field.extraJudges.labelTitle": "Label",
+			"field.extraJudges.labelPlaceholder": "Custom label (optional)",
+			"field.extraJudges.providerAria": "Extra judge #{index} provider",
+			"field.extraJudges.modelAria": "Extra judge #{index} model",
+			"field.extraJudges.effortAria": "Extra judge #{index} reasoning effort",
+			"field.extraJudges.labelAria": "Extra judge #{index} label",
+			"field.extraJudges.removeAria": "Remove extra judge #{index}",
+			"field.extraJudges.conflictPrimary": "Extra judge #{index} duplicates primary judge ({id}). Please select a different model.",
+			"field.extraJudges.conflictDuplicate": "Extra judge #{index} duplicates extra judge #{other} ({id}). Please select a different model.",
 			"section.execution": "Execution",
 			"field.maxConcurrency.title": "Max Concurrency",
 			"field.maxConcurrency.help": "Maximum concurrent requests shared across all verifier tools.",
 			"field.maxRetries.title": "Max Retries",
 			"field.maxRetries.help": "Retry attempts for rate limits, timeouts, and transient network errors.",
+			"field.retryBaseDelayMs.title": "Retry Base Delay",
+			"field.retryBaseDelayMs.help": "Base wait time before retrying on rate limits or transient network errors, in milliseconds.",
 			"field.timeoutMs.title": "Request Timeout",
 			"field.timeoutMs.help": "Timeout for a single model request, in milliseconds.",
+			"field.cacheDir.title": "Cache Relative Directory",
+			"field.cacheDir.help": "Relative directory for persistent score cache (relative to topic directory). Cannot be empty.",
 			"field.cacheMaxEntries.title": "Max Cache Entries",
 			"field.cacheMaxEntries.help": "Maximum entries retained in the local persistent score cache.",
 			"section.cost": "Cost Estimation",
@@ -237,6 +307,24 @@ window.__ModuleLoader__.load({
 			"stats.successCalls": "Successful",
 			"stats.failedCalls": "Failed",
 			"stats.avgDuration": "Avg Duration",
+			"stats.phase.explicit": "Explicit",
+			"stats.phase.compare": "Compare",
+			"stats.phase.select": "Select",
+			"stats.phase.track": "Track",
+			"stats.phase.final": "Final Acceptance",
+			"stats.phase.plan_review": "Plan Review",
+			"stats.phase.team_task": "Team Task",
+			"stats.phase.semantic": "Semantic Route",
+			"stats.outcome.passed": "Passed",
+			"stats.outcome.below-threshold": "Below Threshold",
+			"stats.outcome.tie": "Tie",
+			"stats.outcome.error": "Error",
+			"stats.outcome.dropped-over-budget": "Dropped (over budget)",
+			"stats.outcome.invalid-references": "Invalid references",
+			"stats.winner.tie": "Tie",
+			"stats.verdict.winner": "Winner {winner}",
+			"stats.verdict.scoreThreshold": "Score {score} (threshold {threshold})",
+			"stats.verdict.scoreOnly": "Score {score}",
 			"metric.cacheHitRate": "Cache Hit Rate",
 			"metric.cacheHitNote": "{hits} hits / {total} scored",
 			"metric.tokens": "Tokens",
@@ -355,6 +443,174 @@ window.__ModuleLoader__.load({
 				second: "2-digit"
 			}).format(value);
 		}
+		function resolveCacheDirOnSave(draft, previous) {
+			if (typeof draft === "string" && draft.trim()) return draft.trim();
+			if (typeof previous === "string" && previous.trim()) return previous.trim();
+		}
+		/** An eight-candidate select: ring + pivot rounds (18 pairs) x three criteria, one repeat. */
+		const WORST_CASE_ROUTE_CALLS_PER_JUDGE = 54;
+		/** Final acceptance: three criteria x the default two repeats (one per A/B position). */
+		const WORST_CASE_FINAL_CALLS_PER_JUDGE = 6;
+		const WORST_CASE_TASK_PER_JUDGE = 60;
+		const WORST_CASE_SESSION_PER_JUDGE = 160;
+		function computeJudgeCount(extraJudgesCount) {
+			return 1 + Math.max(0, extraJudgesCount);
+		}
+		function computeWorstCaseBudget(judgeCount) {
+			const count = Math.max(1, judgeCount);
+			return {
+				worstCaseTask: count * 60,
+				worstCaseSession: count * 160
+			};
+		}
+		function evaluateBudgetWarning(autoVerifyMode, extraJudgesCount, autoMaxModelCallsPerTask, autoMaxModelCallsPerSession) {
+			if (autoVerifyMode === "manual") return null;
+			const judgeCount = computeJudgeCount(extraJudgesCount);
+			const { worstCaseTask, worstCaseSession } = computeWorstCaseBudget(judgeCount);
+			const warnTask = autoMaxModelCallsPerTask < worstCaseTask;
+			const warnSession = autoMaxModelCallsPerSession < worstCaseSession;
+			if (!warnTask && !warnSession) return null;
+			return {
+				judgeCount,
+				worstCaseTask,
+				worstCaseSession,
+				warnTask,
+				warnSession
+			};
+		}
+		function isVerdictFailed(verdict, success = true) {
+			if (!success) return true;
+			if (!verdict) return false;
+			return verdict.outcome === "below-threshold" || verdict.outcome === "error";
+		}
+		function formatPercentage(val) {
+			if (!Number.isFinite(val)) return "--";
+			return `${(val * 100).toFixed(1)}%`;
+		}
+		function formatVerdictDetails(verdict, t) {
+			const isFailed = verdict.outcome === "below-threshold" || verdict.outcome === "error";
+			const outcomeKey = verdict.outcome ? `stats.outcome.${verdict.outcome}` : void 0;
+			const outcomeText = outcomeKey && t[outcomeKey] ? t[outcomeKey] : verdict.outcome;
+			const phaseKey = verdict.phase ? `stats.phase.${verdict.phase}` : void 0;
+			const phaseText = phaseKey && t[phaseKey] ? t[phaseKey] : verdict.phase;
+			let scoreText;
+			if (typeof verdict.score === "number") {
+				const scorePct = formatPercentage(verdict.score);
+				if (typeof verdict.threshold === "number") {
+					const threshPct = formatPercentage(verdict.threshold);
+					scoreText = tFormat(t["stats.verdict.scoreThreshold"], {
+						score: scorePct,
+						threshold: threshPct
+					});
+				} else scoreText = tFormat(t["stats.verdict.scoreOnly"], { score: scorePct });
+			}
+			let winnerText;
+			if (verdict.winner) {
+				const winnerValue = verdict.winner === "tie" ? t["stats.winner.tie"] : verdict.winner;
+				winnerText = tFormat(t["stats.verdict.winner"], { winner: winnerValue });
+			}
+			return {
+				outcomeText,
+				phaseText,
+				scoreText,
+				winnerText,
+				isFailed
+			};
+		}
+		function normalizeExtraJudges(value) {
+			if (!Array.isArray(value)) return [];
+			const results = [];
+			for (const item of value) {
+				if (!item || typeof item !== "object" || Array.isArray(item)) continue;
+				const raw = item;
+				if (typeof raw.provider !== "string" || typeof raw.model !== "string") continue;
+				const provider = raw.provider.trim();
+				const model = raw.model.trim();
+				if (!provider || !model) continue;
+				const draft = {
+					provider,
+					model
+				};
+				if (typeof raw.reasoningEffort === "string") {
+					const effort = raw.reasoningEffort.trim();
+					if (effort) draft.reasoningEffort = effort;
+				}
+				if (typeof raw.label === "string") {
+					const label = raw.label.trim();
+					if (label) draft.label = label;
+				}
+				results.push(draft);
+				if (results.length >= 4) break;
+			}
+			return results;
+		}
+		function judgeIdentity(provider, model) {
+			return `${provider.trim()}/${model.trim()}`;
+		}
+		function judgeConflict(primary, judges) {
+			const primaryId = primary.provider.trim() && primary.model.trim() ? judgeIdentity(primary.provider, primary.model) : void 0;
+			const seen = /* @__PURE__ */ new Map();
+			for (let i = 0; i < judges.length; i++) {
+				const judge = judges[i];
+				const p = judge.provider.trim();
+				const m = judge.model.trim();
+				if (!p || !m) continue;
+				const id = judgeIdentity(p, m);
+				if (primaryId !== void 0 && id === primaryId) return {
+					index: i,
+					duplicateOf: "primary"
+				};
+				const prev = seen.get(id);
+				if (prev !== void 0) return {
+					index: i,
+					duplicateOf: prev
+				};
+				seen.set(id, i);
+			}
+		}
+		function addExtraJudge(judges, draft) {
+			if (judges.length >= 4) return [...judges];
+			const item = {
+				provider: draft.provider.trim(),
+				model: draft.model.trim()
+			};
+			if (typeof draft.reasoningEffort === "string") {
+				const effort = draft.reasoningEffort.trim();
+				if (effort) item.reasoningEffort = effort;
+			}
+			if (typeof draft.label === "string") {
+				const label = draft.label.trim();
+				if (label) item.label = label;
+			}
+			return [...judges, item];
+		}
+		function removeExtraJudge(judges, index) {
+			if (index < 0 || index >= judges.length) return [...judges];
+			return judges.filter((_, i) => i !== index);
+		}
+		function serializeExtraJudges(judges) {
+			const list = judges.slice(0, 4);
+			const result = [];
+			for (const judge of list) {
+				const provider = judge.provider.trim();
+				const model = judge.model.trim();
+				if (!provider || !model) continue;
+				const entry = {
+					provider,
+					model
+				};
+				if (typeof judge.reasoningEffort === "string") {
+					const effort = judge.reasoningEffort.trim();
+					if (effort) entry.reasoningEffort = effort;
+				}
+				if (typeof judge.label === "string") {
+					const label = judge.label.trim();
+					if (label) entry.label = label;
+				}
+				result.push(entry);
+			}
+			return result;
+		}
 		//#endregion
 		//#region src/client.tsx
 		const NS = "llm-verifier";
@@ -468,6 +724,7 @@ window.__ModuleLoader__.load({
 				autoVerifyMode: mode,
 				autoVerifyThreshold: Number(v.autoVerifyThreshold ?? .65),
 				autoVerifyRepeats: Number(v.autoVerifyRepeats ?? 1),
+				autoVerifyFinalRepeats: Number(v.autoVerifyFinalRepeats ?? 2),
 				autoVerifyMinToolCalls: Number(v.autoVerifyMinToolCalls ?? 3),
 				autoVerifyMaxChars: Number(v.autoVerifyMaxChars ?? 8e4),
 				autoVerifyMaxPerTask: Number(v.autoVerifyMaxPerTask ?? 2),
@@ -480,7 +737,7 @@ window.__ModuleLoader__.load({
 				autoTrackCompletionThreshold: Number(v.autoTrackCompletionThreshold ?? .8),
 				autoRouteMaxItemChars: Number(v.autoRouteMaxItemChars ?? 2e4),
 				autoRouteMaxInputChars: Number(v.autoRouteMaxInputChars ?? 6e4),
-				autoMaxModelCallsPerTask: Number(v.autoMaxModelCallsPerTask ?? 64),
+				autoMaxModelCallsPerTask: Number(v.autoMaxModelCallsPerTask ?? 96),
 				autoMaxModelCallsPerSession: Number(v.autoMaxModelCallsPerSession ?? 240),
 				autoVerifyTeamTasks: v.autoVerifyTeamTasks !== false,
 				autoVerifyPlanMode: v.autoVerifyPlanMode !== false,
@@ -488,13 +745,18 @@ window.__ModuleLoader__.load({
 				model: String(v.model ?? ""),
 				...typeof v.reasoningEffort === "string" ? { reasoningEffort: v.reasoningEffort } : {},
 				maxTokens: Number(v.maxTokens ?? 32768),
+				temperature: Number(v.temperature ?? .2),
+				...typeof v.label === "string" && v.label.trim() ? { label: v.label.trim() } : {},
 				maxConcurrency: Number(v.maxConcurrency ?? 8),
 				maxRetries: Number(v.maxRetries ?? 3),
+				retryBaseDelayMs: Number(v.retryBaseDelayMs ?? 500),
 				timeoutMs: Number(v.timeoutMs ?? 3e5),
+				cacheDir: typeof v.cacheDir === "string" && v.cacheDir.trim() ? v.cacheDir.trim() : "verifier",
 				cacheMaxEntries: Number(v.cacheMaxEntries ?? 1e4),
 				estimatedInputUsdPerMillion: Number(v.estimatedInputUsdPerMillion ?? 0),
 				estimatedOutputUsdPerMillion: Number(v.estimatedOutputUsdPerMillion ?? 0),
-				autoVerifySubagents: v.autoVerifySubagents === true
+				autoVerifySubagents: v.autoVerifySubagents === true,
+				extraJudges: normalizeExtraJudges(v.extraJudges)
 			};
 		}
 		function message(error) {
@@ -583,17 +845,66 @@ window.__ModuleLoader__.load({
 					[key]: value
 				} : v);
 			};
+			const conflict = (0, react.useMemo)(() => draft ? judgeConflict({
+				provider: draft.provider,
+				model: draft.model
+			}, draft.extraJudges) : void 0, [
+				draft?.provider,
+				draft?.model,
+				draft?.extraJudges
+			]);
+			const addJudge = () => {
+				if (!loaded || !draft || draft.extraJudges.length >= 4) return;
+				const firstGroup = loaded.groups[0];
+				const firstProvider = firstGroup?.id ?? "";
+				const firstModelObj = firstGroup?.models[0];
+				const firstModel = firstModelObj?.id ?? "";
+				const defaultEffort = firstModelObj?.reasoning?.defaultEffort;
+				const newJudge = {
+					provider: firstProvider,
+					model: firstModel,
+					...defaultEffort ? { reasoningEffort: defaultEffort } : {}
+				};
+				patch("extraJudges", addExtraJudge(draft.extraJudges, newJudge));
+			};
+			const removeJudge = (idx) => {
+				if (!draft) return;
+				patch("extraJudges", removeExtraJudge(draft.extraJudges, idx));
+			};
+			const updateJudge = (idx, updates) => {
+				if (!draft) return;
+				const next = draft.extraJudges.map((j, i) => i === idx ? {
+					...j,
+					...updates
+				} : j);
+				patch("extraJudges", next);
+			};
+			const budgetWarning = (0, react.useMemo)(() => {
+				if (!draft) return null;
+				return evaluateBudgetWarning(draft.autoVerifyMode, draft.extraJudges.length, draft.autoMaxModelCallsPerTask, draft.autoMaxModelCallsPerSession);
+			}, [
+				draft?.autoVerifyMode,
+				draft?.extraJudges.length,
+				draft?.autoMaxModelCallsPerTask,
+				draft?.autoMaxModelCallsPerSession
+			]);
 			const save = async () => {
-				if (!loaded || !draft) return;
+				if (!loaded || !draft || conflict) return;
 				setBusy(true);
 				setSaved(false);
 				setError(null);
 				try {
 					const section = {
 						...record(loaded.settings.user),
-						...draft
+						...draft,
+						extraJudges: serializeExtraJudges(draft.extraJudges)
 					};
 					if (!draft.reasoningEffort) delete section.reasoningEffort;
+					if (!draft.label || !draft.label.trim()) delete section.label;
+					else section.label = draft.label.trim();
+					const resolvedCacheDir = resolveCacheDirOnSave(draft.cacheDir, values(loaded.settings).cacheDir);
+					if (!resolvedCacheDir) delete section.cacheDir;
+					else section.cacheDir = resolvedCacheDir;
 					const res = await remote.settings.update(NS, section, loaded.settings.revision);
 					if (!res.ok) throw new Error(res.error.message);
 					setLoaded((v) => v ? {
@@ -654,6 +965,19 @@ window.__ModuleLoader__.load({
 					return next;
 				})
 			});
+			const textField = (key, placeholder) => /* @__PURE__ */ (0, react_jsx_runtime.jsx)(_deepseek_ai_dsh_client_ui_primitives.Input, {
+				style: {
+					width: "100%",
+					height: 36,
+					borderRadius: 8
+				},
+				type: "text",
+				disabled: busy,
+				placeholder,
+				"aria-label": t["field." + key + ".title"] ?? String(key),
+				value: String(draft[key] ?? ""),
+				onChange: (e) => patch(key, e.target.value)
+			});
 			return /* @__PURE__ */ (0, react_jsx_runtime.jsxs)("div", {
 				style: shell,
 				children: [
@@ -842,6 +1166,19 @@ window.__ModuleLoader__.load({
 									help: t["field.autoMaxModelCallsPerTask.help"]
 								}), numeric("autoMaxModelCallsPerTask", 1)]
 							}),
+							budgetWarning?.warnTask && /* @__PURE__ */ (0, react_jsx_runtime.jsx)("p", {
+								style: {
+									margin: "-4px 0 10px",
+									fontSize: 12,
+									lineHeight: "18px",
+									color: "var(--dsw-alias-state-warn-label)"
+								},
+								children: tFormat(t["field.autoMaxModelCallsPerTask.warnBudget"], {
+									current: draft.autoMaxModelCallsPerTask,
+									required: budgetWarning.worstCaseTask,
+									judges: budgetWarning.judgeCount
+								})
+							}),
 							/* @__PURE__ */ (0, react_jsx_runtime.jsxs)("div", {
 								style: row,
 								children: [/* @__PURE__ */ (0, react_jsx_runtime.jsx)(Label, {
@@ -849,6 +1186,19 @@ window.__ModuleLoader__.load({
 									help: t["field.autoMaxModelCallsPerSession.help"]
 								}), numeric("autoMaxModelCallsPerSession", 1)]
 							}),
+							budgetWarning?.warnSession && /* @__PURE__ */ (0, react_jsx_runtime.jsx)("p", {
+								style: {
+									margin: "-4px 0 10px",
+									fontSize: 12,
+									lineHeight: "18px",
+									color: "var(--dsw-alias-state-warn-label)"
+								},
+								children: tFormat(t["field.autoMaxModelCallsPerSession.warnBudget"], {
+									current: draft.autoMaxModelCallsPerSession,
+									required: budgetWarning.worstCaseSession,
+									judges: budgetWarning.judgeCount
+								})
+							}),
 							/* @__PURE__ */ (0, react_jsx_runtime.jsxs)("div", {
 								style: row,
 								children: [/* @__PURE__ */ (0, react_jsx_runtime.jsx)(Label, {
@@ -863,6 +1213,13 @@ window.__ModuleLoader__.load({
 									help: t["field.autoVerifyRepeats.help"]
 								}), numeric("autoVerifyRepeats", 1)]
 							}),
+							/* @__PURE__ */ (0, react_jsx_runtime.jsxs)("div", {
+								style: row,
+								children: [/* @__PURE__ */ (0, react_jsx_runtime.jsx)(Label, {
+									title: t["field.autoVerifyFinalRepeats.title"],
+									help: t["field.autoVerifyFinalRepeats.help"]
+								}), numeric("autoVerifyFinalRepeats", 1)]
+							}),
 							/* @__PURE__ */ (0, react_jsx_runtime.jsxs)("div", {
 								style: row,
 								children: [/* @__PURE__ */ (0, react_jsx_runtime.jsx)(Label, {
@@ -993,6 +1350,188 @@ window.__ModuleLoader__.load({
 									title: t["field.maxTokens.title"],
 									help: t["field.maxTokens.help"]
 								}), numeric("maxTokens", 1)]
+							}),
+							/* @__PURE__ */ (0, react_jsx_runtime.jsxs)("div", {
+								style: row,
+								children: [/* @__PURE__ */ (0, react_jsx_runtime.jsx)(Label, {
+									title: t["field.temperature.title"],
+									help: t["field.temperature.help"]
+								}), numeric("temperature", 0)]
+							}),
+							/* @__PURE__ */ (0, react_jsx_runtime.jsxs)("div", {
+								style: row,
+								children: [/* @__PURE__ */ (0, react_jsx_runtime.jsx)(Label, {
+									title: t["field.label.title"],
+									help: t["field.label.help"]
+								}), /* @__PURE__ */ (0, react_jsx_runtime.jsx)(_deepseek_ai_dsh_client_ui_primitives.Input, {
+									style: {
+										width: "100%",
+										height: 36,
+										borderRadius: 8
+									},
+									type: "text",
+									disabled: busy,
+									placeholder: draft.model || t["field.label.placeholder"],
+									"aria-label": t["field.label.title"],
+									value: draft.label ?? "",
+									onChange: (e) => patch("label", e.target.value || void 0)
+								})]
+							}),
+							/* @__PURE__ */ (0, react_jsx_runtime.jsxs)("div", {
+								style: {
+									...row,
+									gridTemplateColumns: "minmax(180px, 1fr) auto",
+									minHeight: 48
+								},
+								children: [/* @__PURE__ */ (0, react_jsx_runtime.jsx)(Label, {
+									title: t["field.extraJudges.title"],
+									help: t["field.extraJudges.help"]
+								}), /* @__PURE__ */ (0, react_jsx_runtime.jsx)(_deepseek_ai_dsh_client_ui_primitives.Button, {
+									variant: "outline",
+									disabled: busy || draft.extraJudges.length >= 4,
+									onClick: addJudge,
+									children: t["field.extraJudges.add"]
+								})]
+							}),
+							draft.extraJudges.length > 0 && /* @__PURE__ */ (0, react_jsx_runtime.jsxs)("div", {
+								style: {
+									display: "flex",
+									flexDirection: "column",
+									gap: 8,
+									padding: "10px 0",
+									borderBottom: "1px solid var(--dsw-alias-border-l2)"
+								},
+								children: [/* @__PURE__ */ (0, react_jsx_runtime.jsxs)("div", {
+									style: {
+										display: "grid",
+										gridTemplateColumns: "minmax(110px, 1.2fr) minmax(120px, 1.3fr) minmax(95px, 1fr) minmax(95px, 1fr) auto",
+										gap: 8,
+										fontSize: 12,
+										color: "var(--dsw-alias-label-tertiary)",
+										paddingBottom: 2
+									},
+									children: [
+										/* @__PURE__ */ (0, react_jsx_runtime.jsx)("div", { children: t["field.provider.title"] }),
+										/* @__PURE__ */ (0, react_jsx_runtime.jsx)("div", { children: t["field.model.title"] }),
+										/* @__PURE__ */ (0, react_jsx_runtime.jsx)("div", { children: t["field.reasoningEffort.title"] }),
+										/* @__PURE__ */ (0, react_jsx_runtime.jsx)("div", { children: t["field.extraJudges.labelTitle"] }),
+										/* @__PURE__ */ (0, react_jsx_runtime.jsx)("div", {})
+									]
+								}), draft.extraJudges.map((judge, idx) => {
+									const judgeModels = loaded.groups.find((g) => g.id === judge.provider)?.models ?? [];
+									const judgeEfforts = judgeModels.find((m) => m.id === judge.model)?.reasoning?.efforts ?? [];
+									return /* @__PURE__ */ (0, react_jsx_runtime.jsxs)("div", {
+										style: {
+											display: "grid",
+											gridTemplateColumns: "minmax(110px, 1.2fr) minmax(120px, 1.3fr) minmax(95px, 1fr) minmax(95px, 1fr) auto",
+											gap: 8,
+											alignItems: "center"
+										},
+										children: [
+											/* @__PURE__ */ (0, react_jsx_runtime.jsxs)("select", {
+												style: selectStyle,
+												disabled: busy,
+												"aria-label": tFormat(t["field.extraJudges.providerAria"], { index: idx + 1 }),
+												value: judge.provider,
+												onChange: (e) => {
+													const newProvider = e.target.value;
+													const firstM = loaded.groups.find((g) => g.id === newProvider)?.models[0];
+													updateJudge(idx, {
+														provider: newProvider,
+														model: firstM?.id ?? "",
+														...firstM?.reasoning?.defaultEffort ? { reasoningEffort: firstM.reasoning.defaultEffort } : { reasoningEffort: void 0 }
+													});
+												},
+												children: [!loaded.groups.some((g) => g.id === judge.provider) && judge.provider && /* @__PURE__ */ (0, react_jsx_runtime.jsx)("option", {
+													value: judge.provider,
+													children: judge.provider
+												}), loaded.groups.map((g) => /* @__PURE__ */ (0, react_jsx_runtime.jsxs)("option", {
+													value: g.id,
+													children: [
+														g.name,
+														" · ",
+														g.id
+													]
+												}, g.id))]
+											}),
+											/* @__PURE__ */ (0, react_jsx_runtime.jsxs)("select", {
+												style: selectStyle,
+												disabled: busy,
+												"aria-label": tFormat(t["field.extraJudges.modelAria"], { index: idx + 1 }),
+												value: judge.model,
+												onChange: (e) => {
+													const newModel = e.target.value;
+													const found = judgeModels.find((m) => m.id === newModel);
+													updateJudge(idx, {
+														model: newModel,
+														...found?.reasoning?.defaultEffort ? { reasoningEffort: found.reasoning.defaultEffort } : { reasoningEffort: void 0 }
+													});
+												},
+												children: [!judgeModels.some((m) => m.id === judge.model) && judge.model && /* @__PURE__ */ (0, react_jsx_runtime.jsx)("option", {
+													value: judge.model,
+													children: judge.model
+												}), judgeModels.map((m) => /* @__PURE__ */ (0, react_jsx_runtime.jsxs)("option", {
+													value: m.id,
+													children: [
+														m.name,
+														" · ",
+														m.id
+													]
+												}, m.id))]
+											}),
+											/* @__PURE__ */ (0, react_jsx_runtime.jsxs)("select", {
+												style: selectStyle,
+												disabled: busy,
+												"aria-label": tFormat(t["field.extraJudges.effortAria"], { index: idx + 1 }),
+												value: judge.reasoningEffort ?? "",
+												onChange: (e) => updateJudge(idx, { reasoningEffort: e.target.value || void 0 }),
+												children: [/* @__PURE__ */ (0, react_jsx_runtime.jsx)("option", {
+													value: "",
+													children: t["field.reasoningEffort.default"]
+												}), judgeEfforts.map((ef) => /* @__PURE__ */ (0, react_jsx_runtime.jsx)("option", {
+													value: ef.id,
+													children: ef.name
+												}, ef.id))]
+											}),
+											/* @__PURE__ */ (0, react_jsx_runtime.jsx)(_deepseek_ai_dsh_client_ui_primitives.Input, {
+												style: {
+													width: "100%",
+													height: 36,
+													borderRadius: 8
+												},
+												type: "text",
+												disabled: busy,
+												placeholder: t["field.extraJudges.labelPlaceholder"],
+												"aria-label": tFormat(t["field.extraJudges.labelAria"], { index: idx + 1 }),
+												value: judge.label ?? "",
+												onChange: (e) => updateJudge(idx, { label: e.target.value })
+											}),
+											/* @__PURE__ */ (0, react_jsx_runtime.jsx)(_deepseek_ai_dsh_client_ui_primitives.Button, {
+												variant: "outline",
+												disabled: busy,
+												"aria-label": tFormat(t["field.extraJudges.removeAria"], { index: idx + 1 }),
+												onClick: () => removeJudge(idx),
+												children: t["field.extraJudges.remove"]
+											})
+										]
+									}, idx);
+								})]
+							}),
+							conflict && /* @__PURE__ */ (0, react_jsx_runtime.jsx)("p", {
+								style: {
+									margin: "8px 0 0",
+									fontSize: 12,
+									lineHeight: "18px",
+									color: "var(--dsw-alias-state-warn-label)"
+								},
+								children: conflict.duplicateOf === "primary" ? tFormat(t["field.extraJudges.conflictPrimary"], {
+									index: conflict.index + 1,
+									id: judgeIdentity(draft.extraJudges[conflict.index]?.provider ?? "", draft.extraJudges[conflict.index]?.model ?? "")
+								}) : tFormat(t["field.extraJudges.conflictDuplicate"], {
+									index: conflict.index + 1,
+									other: conflict.duplicateOf + 1,
+									id: judgeIdentity(draft.extraJudges[conflict.index]?.provider ?? "", draft.extraJudges[conflict.index]?.model ?? "")
+								})
 							})
 						]
 					}),
@@ -1014,6 +1553,13 @@ window.__ModuleLoader__.load({
 									help: t["field.maxRetries.help"]
 								}), numeric("maxRetries", 0)]
 							}),
+							/* @__PURE__ */ (0, react_jsx_runtime.jsxs)("div", {
+								style: row,
+								children: [/* @__PURE__ */ (0, react_jsx_runtime.jsx)(Label, {
+									title: t["field.retryBaseDelayMs.title"],
+									help: t["field.retryBaseDelayMs.help"]
+								}), numeric("retryBaseDelayMs", 1)]
+							}),
 							/* @__PURE__ */ (0, react_jsx_runtime.jsxs)("div", {
 								style: row,
 								children: [/* @__PURE__ */ (0, react_jsx_runtime.jsx)(Label, {
@@ -1021,6 +1567,13 @@ window.__ModuleLoader__.load({
 									help: t["field.timeoutMs.help"]
 								}), numeric("timeoutMs", 1)]
 							}),
+							/* @__PURE__ */ (0, react_jsx_runtime.jsxs)("div", {
+								style: row,
+								children: [/* @__PURE__ */ (0, react_jsx_runtime.jsx)(Label, {
+									title: t["field.cacheDir.title"],
+									help: t["field.cacheDir.help"]
+								}), textField("cacheDir", "verifier")]
+							}),
 							/* @__PURE__ */ (0, react_jsx_runtime.jsxs)("div", {
 								style: row,
 								children: [/* @__PURE__ */ (0, react_jsx_runtime.jsx)(Label, {
@@ -1099,7 +1652,7 @@ window.__ModuleLoader__.load({
 							children: t["settings.reload"]
 						}), /* @__PURE__ */ (0, react_jsx_runtime.jsx)(_deepseek_ai_dsh_client_ui_primitives.Button, {
 							variant: "primary",
-							disabled: busy || !loaded.writable,
+							disabled: busy || !loaded.writable || Boolean(conflict),
 							onClick: () => void save(),
 							children: busy ? t["settings.saving"] : t["settings.save"]
 						})]
@@ -1768,65 +2321,113 @@ window.__ModuleLoader__.load({
 											maxHeight: 360,
 											overflow: "auto"
 										},
-										children: [(data?.recent ?? []).map((item) => /* @__PURE__ */ (0, react_jsx_runtime.jsxs)("div", {
-											style: {
-												display: "grid",
-												gridTemplateColumns: "minmax(170px,1fr) auto",
-												gap: 12,
-												padding: "11px 8px",
-												borderTop: "1px solid var(--dsw-alias-border-l2, rgba(255,255,255,.1))"
-											},
-											children: [/* @__PURE__ */ (0, react_jsx_runtime.jsxs)("div", { children: [/* @__PURE__ */ (0, react_jsx_runtime.jsxs)("div", {
+										children: [(data?.recent ?? []).map((item) => {
+											const verdictInfo = item.verdict ? formatVerdictDetails(item.verdict, t) : void 0;
+											return /* @__PURE__ */ (0, react_jsx_runtime.jsxs)("div", {
 												style: {
-													display: "flex",
-													alignItems: "center",
-													gap: 8
+													display: "grid",
+													gridTemplateColumns: "minmax(170px,1fr) auto",
+													gap: 12,
+													padding: "11px 8px",
+													borderTop: "1px solid var(--dsw-alias-border-l2, rgba(255,255,255,.1))"
 												},
-												children: [
-													/* @__PURE__ */ (0, react_jsx_runtime.jsx)("span", { style: {
-														width: 7,
-														height: 7,
-														borderRadius: "50%",
-														background: item.success ? "#59c985" : "#e76565"
-													} }),
-													/* @__PURE__ */ (0, react_jsx_runtime.jsx)("strong", {
-														style: { fontSize: 13 },
-														children: labels[item.toolName] ?? item.toolName
+												children: [/* @__PURE__ */ (0, react_jsx_runtime.jsxs)("div", { children: [
+													/* @__PURE__ */ (0, react_jsx_runtime.jsxs)("div", {
+														style: {
+															display: "flex",
+															alignItems: "center",
+															gap: 8
+														},
+														children: [
+															/* @__PURE__ */ (0, react_jsx_runtime.jsx)("span", { style: {
+																width: 7,
+																height: 7,
+																borderRadius: "50%",
+																background: !item.success || (verdictInfo ? verdictInfo.isFailed : false) ? "#e76565" : "#59c985"
+															} }),
+															/* @__PURE__ */ (0, react_jsx_runtime.jsx)("strong", {
+																style: { fontSize: 13 },
+																children: labels[item.toolName] ?? item.toolName
+															}),
+															/* @__PURE__ */ (0, react_jsx_runtime.jsxs)("span", {
+																style: muted,
+																children: [
+																	item.provider,
+																	"/",
+																	item.model
+																]
+															})
+														]
 													}),
-													/* @__PURE__ */ (0, react_jsx_runtime.jsxs)("span", {
-														style: muted,
+													!item.success && /* @__PURE__ */ (0, react_jsx_runtime.jsx)("div", {
+														title: item.errorMessage,
+														style: {
+															margin: "5px 0 0 15px",
+															fontSize: 11,
+															color: "#e76565",
+															overflow: "hidden",
+															textOverflow: "ellipsis",
+															whiteSpace: "nowrap"
+														},
+														children: item.errorMessage ?? item.errorName
+													}),
+													item.verdict && verdictInfo && /* @__PURE__ */ (0, react_jsx_runtime.jsxs)("div", {
+														style: {
+															display: "flex",
+															alignItems: "center",
+															flexWrap: "wrap",
+															gap: 6,
+															margin: "6px 0 0 15px",
+															fontSize: 11
+														},
 														children: [
-															item.provider,
-															"/",
-															item.model
+															verdictInfo.outcomeText && /* @__PURE__ */ (0, react_jsx_runtime.jsx)("span", {
+																style: {
+																	padding: "1px 5px",
+																	borderRadius: 4,
+																	fontWeight: 500,
+																	fontSize: 11,
+																	background: verdictInfo.isFailed ? "rgba(231,101,101,.16)" : item.verdict.outcome === "tie" ? "rgba(227,189,99,.16)" : "rgba(89,201,133,.16)",
+																	color: verdictInfo.isFailed ? "#e76565" : item.verdict.outcome === "tie" ? "#e3bd63" : "#77d49b",
+																	border: `1px solid ${verdictInfo.isFailed ? "rgba(231,101,101,.3)" : item.verdict.outcome === "tie" ? "rgba(227,189,99,.3)" : "rgba(89,201,133,.3)"}`
+																},
+																children: verdictInfo.outcomeText
+															}),
+															verdictInfo.phaseText && /* @__PURE__ */ (0, react_jsx_runtime.jsx)("span", {
+																style: {
+																	padding: "1px 5px",
+																	borderRadius: 4,
+																	background: "var(--dsw-surface-sunken)",
+																	color: "var(--dsw-text-secondary)",
+																	border: "1px solid var(--dsw-alias-border-l2, rgba(255,255,255,.1))"
+																},
+																children: verdictInfo.phaseText
+															}),
+															verdictInfo.scoreText && /* @__PURE__ */ (0, react_jsx_runtime.jsx)("span", {
+																style: { color: typeof item.verdict.threshold === "number" && typeof item.verdict.score === "number" && item.verdict.score < item.verdict.threshold ? "#e76565" : "var(--dsw-text-primary)" },
+																children: verdictInfo.scoreText
+															}),
+															verdictInfo.winnerText && /* @__PURE__ */ (0, react_jsx_runtime.jsx)("span", {
+																style: { color: "var(--dsw-text-secondary)" },
+																children: verdictInfo.winnerText
+															})
 														]
 													})
-												]
-											}), !item.success && /* @__PURE__ */ (0, react_jsx_runtime.jsx)("div", {
-												title: item.errorMessage,
-												style: {
-													margin: "5px 0 0 15px",
-													fontSize: 11,
-													color: "#e76565",
-													overflow: "hidden",
-													textOverflow: "ellipsis",
-													whiteSpace: "nowrap"
-												},
-												children: item.errorMessage ?? item.errorName
-											})] }), /* @__PURE__ */ (0, react_jsx_runtime.jsxs)("div", {
-												style: { textAlign: "right" },
-												children: [/* @__PURE__ */ (0, react_jsx_runtime.jsx)("div", {
-													style: { fontSize: 12 },
-													children: duration(item.durationMs)
-												}), /* @__PURE__ */ (0, react_jsx_runtime.jsx)("div", {
-													style: {
-														...muted,
-														marginTop: 3
-													},
-													children: dateTime(item.startedAt, lang)
+												] }), /* @__PURE__ */ (0, react_jsx_runtime.jsxs)("div", {
+													style: { textAlign: "right" },
+													children: [/* @__PURE__ */ (0, react_jsx_runtime.jsx)("div", {
+														style: { fontSize: 12 },
+														children: duration(item.durationMs)
+													}), /* @__PURE__ */ (0, react_jsx_runtime.jsx)("div", {
+														style: {
+															...muted,
+															marginTop: 3
+														},
+														children: dateTime(item.startedAt, lang)
+													})]
 												})]
-											})]
-										}, item.id)), (data?.recent.length ?? 0) === 0 && /* @__PURE__ */ (0, react_jsx_runtime.jsx)("div", {
+											}, item.id);
+										}), (data?.recent.length ?? 0) === 0 && /* @__PURE__ */ (0, react_jsx_runtime.jsx)("div", {
 											style: {
 												padding: 24,
 												textAlign: "center",
@@ -2012,15 +2613,26 @@ window.__ModuleLoader__.load({
 		exports.StatisticsPage = StatisticsPage;
 		exports.VerifierSettings = VerifierSettings;
 		exports.VerifierSidebarIcon = VerifierSidebarIcon;
+		exports.WORST_CASE_FINAL_CALLS_PER_JUDGE = WORST_CASE_FINAL_CALLS_PER_JUDGE;
+		exports.WORST_CASE_ROUTE_CALLS_PER_JUDGE = WORST_CASE_ROUTE_CALLS_PER_JUDGE;
+		exports.WORST_CASE_SESSION_PER_JUDGE = WORST_CASE_SESSION_PER_JUDGE;
+		exports.WORST_CASE_TASK_PER_JUDGE = WORST_CASE_TASK_PER_JUDGE;
 		exports.apply = apply;
 		exports.compact = compact;
+		exports.computeJudgeCount = computeJudgeCount;
+		exports.computeWorstCaseBudget = computeWorstCaseBudget;
 		exports.dateTime = dateTime;
 		exports.detectLanguage = detectLanguage;
 		exports.dictionaries = dictionaries;
 		exports.duration = duration;
 		exports.en = en;
+		exports.evaluateBudgetWarning = evaluateBudgetWarning;
+		exports.formatPercentage = formatPercentage;
+		exports.formatVerdictDetails = formatVerdictDetails;
 		exports.inject = inject;
+		exports.isVerdictFailed = isVerdictFailed;
 		exports.money = money;
+		exports.resolveCacheDirOnSave = resolveCacheDirOnSave;
 		exports.tFormat = tFormat;
 		exports.toolLabels = toolLabels;
 		exports.useLanguage = useLanguage;