oh-my-knowledge 0.48.0 → 0.49.0

package/dist/types/observability.d.ts

@@ -11,7 +11,18 @@
  *   - 跨层共享类型集中在 types/，单一来源
  */
 import type { DiagnosisBundle } from './diagnosis.js';
+import type { ToolCallStatus } from './executor.js';
+import type { TraceSourceKind } from './trace.js';
 import type { SkillHardRule, SkillWorkflow } from '../shared/hard-rules.js';
+export interface TraceIngestionSummary {
+    fileCount: number;
+    sourceRecordCount: number;
+    parsedRecordCount: number;
+    malformedRecordCount: number;
+    ignoredValueCount: number;
+    unknownEventCount: number;
+    filteredSessionCount: number;
+}
 export interface TraceSourceMetadata {
     channel?: string;
     sender?: string;
@@ -35,10 +46,12 @@ export interface ObservationReviewStateEntry {
     metricKey?: ObservationMetricKey;
     metricScope?: ObservationMetricScope;
     metricScopeId?: string;
+    traceId?: string;
     sourceTrace?: string;
     sessionId?: string;
     messageIndex?: number;
     messageUuid?: string;
+    callInstanceId?: string;
     toolUseId?: string;
     snippet?: string;
 }
@@ -57,10 +70,12 @@ export interface ObservationReviewStateUpdate {
     metricKey?: ObservationMetricKey;
     metricScope?: ObservationMetricScope;
     metricScopeId?: string;
+    traceId?: string;
     sourceTrace?: string;
     sessionId?: string;
     messageIndex?: number;
     messageUuid?: string;
+    callInstanceId?: string;
     toolUseId?: string;
     snippet?: string;
 }
@@ -69,12 +84,14 @@ export type ExperienceProblemSignal = 'user_correction' | 'negative_feedback' |
 export interface ExperienceProblemEvidenceRef {
     id: string;
     kind: string;
+    traceId?: string;
     sourceTrace: string;
     sessionId: string;
     messageIndex?: number;
     logicalMessageIndex?: number;
     sourceLineIndex?: number;
     messageUuid?: string;
+    callInstanceId?: string;
     toolUseId?: string;
     timestamp?: string;
     role?: 'user' | 'assistant' | 'tool' | 'other';
@@ -95,12 +112,14 @@ export interface ExperienceProblemPattern {
 export interface ProblemTimelineEvent {
     id: string;
     kind: string;
+    traceId?: string;
     sourceTrace: string;
     sessionId: string;
     messageIndex?: number;
     logicalMessageIndex?: number;
     sourceLineIndex?: number;
     messageUuid?: string;
+    callInstanceId?: string;
     toolUseId?: string;
     timestamp?: string;
     role?: 'user' | 'assistant' | 'tool' | 'other';
@@ -119,10 +138,15 @@ export interface SkillChainAdvisory {
     shortLabel: string;
 }
 export type ObservationSignalType = 'failed_search' | 'repeated_failure' | 'hedging' | 'explicit_marker';
-export type ObservationSourceKind = 'claude' | 'openclaw' | 'markdown_log' | 'unknown';
+/** @deprecated Prefer TraceSourceKind for new source-neutral APIs. */
+export type ObservationSourceKind = TraceSourceKind;
 export type ObservationSeverityReasonCode = 'knowledge_gap_suspected' | 'repeated_failure_suspected' | 'explicit_gap_marker' | 'exploratory_probe' | 'skill_asset_unavailable' | 'soft_hedging_signal' | 'tool_or_runtime_noise';
 export type ObservationSignalSubtype = 'hard_miss' | 'repeated_failure' | 'exploratory_miss' | 'tool_error' | 'permission_error' | 'bash_probe' | 'not_found' | 'transient_file_missing' | 'skill_asset_read_failed' | 'permission_denied' | 'tool_limit' | 'tool_failure' | 'regex_only' | 'llm_classified' | 'marker';
 export interface ObservationEvidence {
+    traceId?: string;
+    sessionId?: string;
+    sourceTrace?: string;
+    sourceKind?: ObservationSourceKind;
     tool?: string;
     query?: string;
     path?: string;
@@ -131,6 +155,7 @@ export interface ObservationEvidence {
     markerToken?: string;
     messageIndex?: number;
     messageUuid?: string;
+    callInstanceId?: string;
     toolUseId?: string;
     segmentTimestamp?: string;
 }
@@ -154,6 +179,8 @@ export interface ObservationInboxItem {
     artifactHash?: string;
     cwd?: string;
     sessionId: string;
+    /** Physical evidence stream identity; unlike sessionId, unique across reused run ids. */
+    traceId?: string;
     sourceTrace: string;
     sourceKind: ObservationSourceKind;
     signalType: ObservationSignalType;
@@ -168,11 +195,15 @@ export interface ObservationInboxItem {
     firstSeen: string;
     lastSeen: string;
     occurrences: number;
+    /** Occurrences backed by an observed source timestamp. */
+    timestampedOccurrences?: number;
     recentSessionIds: string[];
+    recentTraceIds?: string[];
     representativeEvidence: ObservationEvidence[];
 }
 export interface ObservationSessionTimeRange {
     sessionId: string;
+    traceId?: string;
     sessionGroupId?: string;
     sourceTrace: string;
     sourceKind: ObservationSourceKind;
@@ -196,12 +227,15 @@ export interface ObservationInboxReport {
             durationMs?: number;
         };
         sessionTimeRanges?: ObservationSessionTimeRange[];
+        ingestion?: TraceIngestionSummary;
         segmentCount: number;
         itemCount: number;
         skillInvocationCounts?: Record<string, number>;
         skillSessionCounts?: Record<string, number>;
         skillInvocationLastSeen?: Record<string, string>;
         skillToolCallCounts?: Record<string, Record<string, number>>;
+        timestampedSegmentCount?: number;
+        timestampCoverage?: number;
     };
     items: ObservationInboxItem[];
     experience?: ObservationExperienceReport;
@@ -254,6 +288,7 @@ export type ExperienceRuleFindingCode = 'high_observation_seen' | 'medium_observ
 export interface ExperienceEvidenceRef {
     id: string;
     kind: ExperienceEvidenceKind;
+    traceId?: string;
     sourceTrace: string;
     sessionId: string;
     traceRole?: 'standalone' | 'main' | 'subagent';
@@ -262,6 +297,8 @@ export interface ExperienceEvidenceRef {
     logicalMessageIndex?: number;
     sourceLineIndex?: number;
     messageUuid?: string;
+    /** Source-neutral identity for one concrete tool-call occurrence. */
+    callInstanceId?: string;
     toolUseId?: string;
     timestamp?: string;
     role?: 'user' | 'assistant' | 'tool' | 'other';
@@ -271,17 +308,22 @@ export interface ExperienceEvidenceRef {
 export interface ExperienceTimelineEvent extends ExperienceEvidenceRef {
     order: number;
     toolName?: string;
+    toolStatus?: ToolCallStatus;
     isError?: boolean;
     fullText?: string;
 }
 export interface ExperienceTimelineBranch {
     id: string;
     label: string;
+    sessionId: string;
+    traceId?: string;
     sourceTrace: string;
     traceRole: 'main' | 'subagent' | 'standalone';
     attachTo?: {
+        traceId?: string;
         sourceTrace: string;
         messageIndex?: number;
+        callInstanceId?: string;
         toolUseId?: string;
         label?: string;
     };
@@ -292,6 +334,20 @@ export interface ExperienceTimelineTree {
     main: ExperienceTimelineEvent[];
     branches: ExperienceTimelineBranch[];
 }
+export interface ExperienceTraceTimeline {
+    id: string;
+    sessionGroupKey: string;
+    sessionId: string;
+    eventCount: number;
+    tree: ExperienceTimelineTree;
+}
+export interface ExperienceTraceRecordRange {
+    traceId: string;
+    sourceTrace: string;
+    startRecordIndex: number;
+    endRecordIndex: number;
+    eventCount: number;
+}
 export interface ExperienceEvidenceChain {
     userMessageCount: number;
     runtimeContextCount: number;
@@ -391,10 +447,13 @@ export interface ExperienceSessionStorySubagentDispatch {
     id: string;
     order: number;
     branchId: string;
+    childSessionId: string;
+    traceId: string;
     label: string;
     sourceTrace: string;
     attachTo?: {
         messageIndex?: number;
+        callInstanceId?: string;
         toolUseId?: string;
         label?: string;
     };
@@ -419,6 +478,7 @@ export interface ExperienceGoalEvidenceRef {
 export interface ExperienceMessageRange {
     startMessageIndex: number;
     endMessageIndex: number;
+    traceId?: string;
     sourceTrace?: string;
     sessionId?: string;
 }
@@ -521,6 +581,7 @@ export interface ExperienceSessionStoryGraphEdge {
 }
 export interface ExperienceSessionStory {
     schemaVersion: 1;
+    contextRef?: string;
     summary: string;
     invocationCount: number;
     goalSliceCount: number;
@@ -539,6 +600,13 @@ export interface ExperienceSessionStory {
     nodes: ExperienceSessionStoryNode[];
     answers: ExperienceSessionStoryAnswer[];
 }
+export interface ExperienceStoryContext {
+    id: string;
+    sessionGroupKey: string;
+    goalSlices: ExperienceSessionStoryGoalSlice[];
+    subagentDispatches: ExperienceSessionStorySubagentDispatch[];
+    episodes: ExperienceEpisode[];
+}
 export interface ExperienceReviewerReport {
     schemaVersion: 1;
     mode: 'deterministic_milestone_1' | 'deterministic_session_story';
@@ -554,6 +622,8 @@ export interface ExperienceReviewerReport {
     oneLookMetrics: {
         toolCallCount: number;
         toolFailureCount: number;
+        toolCancelledCount?: number;
+        toolUnknownCount?: number;
         userMessageCount: number;
         userFollowUpCount: number;
         assistantDeliverySignalCount: number;
@@ -570,10 +640,15 @@ export interface ExperienceReviewerReport {
             outputTokens: number;
             cacheReadTokens: number;
             cacheCreationTokens: number;
+            /** Missing on legacy reviewer reports; readers must treat that as unknown coverage. */
+            observedInvocationCount?: number;
+            invocationCount?: number;
+            coverage?: number;
             attribution: 'skill_segment';
         };
     };
     sessionStory: ExperienceSessionStory;
+    sessionStoryRef?: 'session';
     authorSuggestions: string[];
     traceLinks: ExperienceEvidenceRef[];
 }
@@ -581,10 +656,12 @@ export interface ExperienceGoalSlice {
     id: string;
     skillName: string;
     sessionId: string;
+    traceId?: string;
     sourceTrace: string;
     cwd?: string;
     startTimestamp: string;
     endTimestamp: string;
+    timestampObserved?: boolean;
     sliceReasonCode: ExperienceGoalSliceReasonCode;
     sliceConfidence: 'low' | 'medium' | 'high';
     inferredUserGoal?: string;
@@ -608,6 +685,9 @@ export interface ExperienceReviewIndicators {
     repeatedExecutionCount: number;
     toolCallCount: number;
     toolFailureCount: number;
+    toolCancelledCount?: number;
+    /** Runtime did not expose a trustworthy terminal outcome. */
+    toolUnknownCount?: number;
     highObservationCount: number;
     mediumObservationCount: number;
     hedgingCount: number;
@@ -619,15 +699,21 @@ export interface ExperienceInvocationMetrics {
     outputTokens: number;
     cacheReadTokens: number;
     cacheCreationTokens: number;
+    /** False means counters are placeholders because the trace exposed no valid usage event. */
+    tokenUsageObserved: boolean;
     numTurns: number;
     numToolCalls: number;
     numToolFailures: number;
+    numToolCancelled?: number;
+    /** Unresolved or source-unknown outcomes; excluded from failure-rate denominators. */
+    numToolUnknown?: number;
 }
 export interface ExperienceInvocation {
     id: string;
     skillName: string;
     sessionId: string;
     sessionGroupKey: string;
+    traceId?: string;
     sourceTrace: string;
     sourceKind: ObservationSourceKind;
     entrypoint?: string;
@@ -637,6 +723,7 @@ export interface ExperienceInvocation {
     goalSliceId: string;
     startTimestamp: string;
     endTimestamp: string;
+    timestampObserved?: boolean;
     attribution: {
         source: string;
         confidence: number;
@@ -653,6 +740,8 @@ export interface ExperienceInvocation {
     problemPatterns: ExperienceProblemPattern[];
     relatedObservationIds: string[];
     evidenceRefs: ExperienceEvidenceRef[];
+    timelineRef?: string;
+    timelineEventIds?: string[];
     timeline: ExperienceTimelineEvent[];
 }
 export interface ExperienceSessionSummary {
@@ -669,6 +758,8 @@ export interface ExperienceSessionSummary {
     sourceSessionDurationMs?: number;
     startTimestamp: string;
     endTimestamp: string;
+    timestampedInvocationCount?: number;
+    timestampCoverage?: number;
     invocationIds: string[];
     goalSliceIds: string[];
     reviewPriority: ExperienceReviewPriority;
@@ -680,22 +771,34 @@ export interface ExperienceSessionSummary {
     assistiveInference: ExperienceAssistiveInference;
     problemPatterns: ExperienceProblemPattern[];
     relatedObservationIds: string[];
+    timelineRef?: string;
+    timelinePreviewEventIds?: string[];
     timelinePreview: ExperienceTimelineEvent[];
     fullSessionTimeline: ExperienceTimelineEvent[];
     timelineTree?: ExperienceTimelineTree;
     timelineScope: {
         mode: 'skill_segment_window';
-        segmentStartRecordIndex?: number;
-        segmentEndRecordIndex?: number;
-        previewStartRecordIndex?: number;
-        previewEndRecordIndex?: number;
-        sessionStartRecordIndex: number;
-        sessionEndRecordIndex: number;
+        segmentEventCount: number;
         previewEventCount: number;
         fullSessionEventCount: number;
+        segmentRecordRanges: ExperienceTraceRecordRange[];
+        previewRecordRanges: ExperienceTraceRecordRange[];
+        sessionRecordRanges: ExperienceTraceRecordRange[];
         truncated: boolean;
         omittedBeforeCount: number;
         omittedAfterCount: number;
+        /** @deprecated v2 only. Record indexes are local to one physical trace. */
+        segmentStartRecordIndex?: number;
+        /** @deprecated v2 only. Record indexes are local to one physical trace. */
+        segmentEndRecordIndex?: number;
+        /** @deprecated v2 only. Record indexes are local to one physical trace. */
+        previewStartRecordIndex?: number;
+        /** @deprecated v2 only. Record indexes are local to one physical trace. */
+        previewEndRecordIndex?: number;
+        /** @deprecated v2 only. Record indexes are local to one physical trace. */
+        sessionStartRecordIndex?: number;
+        /** @deprecated v2 only. Record indexes are local to one physical trace. */
+        sessionEndRecordIndex?: number;
     };
     attributionSources: string[];
     pluginNames: string[];
@@ -725,6 +828,8 @@ export interface ExperienceSkillSummary {
     toolCounts: Record<string, number>;
     firstSeen: string;
     lastSeen: string;
+    timestampedInvocationCount?: number;
+    timestampCoverage?: number;
     reviewFirstSessionCount: number;
     sampleReviewSessionCount: number;
     indicators: ExperienceReviewIndicators;
@@ -736,7 +841,7 @@ export interface ExperienceSkillSummary {
 }
 export interface ObservationExperienceReport {
     kind: 'observe-experience';
-    schemaVersion: 2;
+    schemaVersion: 3;
     scope: 'evidence-only';
     generatedAt: string;
     meta: {
@@ -747,6 +852,8 @@ export interface ObservationExperienceReport {
         noteCodes: Array<'no_llm_judge' | 'no_auto_verdict' | 'default_goal_slice_is_allowed' | 'deterministic_assistive_inference'>;
     };
     goalSlices: ExperienceGoalSlice[];
+    traceTimelines: ExperienceTraceTimeline[];
+    storyContexts: ExperienceStoryContext[];
     invocations: ExperienceInvocation[];
     sessions: ExperienceSessionSummary[];
     skills: ExperienceSkillSummary[];
@@ -763,7 +870,7 @@ export interface ObservationRuntimeCheck {
     evidenceSnippets: string[];
 }
 export type SkillRuntimeEvidencePackSourceType = 'tool_call' | 'tool_result' | 'assistant_message' | 'user_feedback' | 'artifact' | 'runtime_context' | 'skill_context' | 'unknown';
-export interface SkillRuntimeEvidencePackRef extends Pick<ExperienceEvidenceRef, 'id' | 'kind' | 'sourceTrace' | 'sessionId' | 'messageUuid' | 'messageIndex' | 'logicalMessageIndex' | 'sourceLineIndex' | 'toolUseId' | 'timestamp' | 'role' | 'label' | 'snippet'> {
+export interface SkillRuntimeEvidencePackRef extends Pick<ExperienceEvidenceRef, 'id' | 'kind' | 'sourceTrace' | 'sessionId' | 'messageUuid' | 'messageIndex' | 'logicalMessageIndex' | 'sourceLineIndex' | 'callInstanceId' | 'toolUseId' | 'timestamp' | 'role' | 'label' | 'snippet'> {
     sourceType: SkillRuntimeEvidencePackSourceType;
     toolName?: string;
     isError?: boolean;

package/dist/types/report.d.ts

@@ -10,11 +10,14 @@ export interface VariantResult {
     totalTokens: number;
     cacheReadTokens: number;
     cacheCreationTokens: number;
+    /** Mirrors `ExecResult.tokenUsageReportedByExecutor`.
+     *  False means token counters are placeholders for unavailable telemetry. */
+    tokenUsageReportedByExecutor?: boolean;
     execCostUSD: number;
     judgeCostUSD: number;
     /** Diagnostic 自身花费(USD)。仅在 failed-assertion 触发 diagnostic 且 executor 报告了 cost 时有值。
-     *  Diagnostic 一般跑在 judge executor 上(claude:haiku 标配),所以 cost-reported 语义随
-     *  `judgeCostReportedByExecutor`,不再单独引一个 flag。
+     *  Diagnostic 跟随首位 judge executor，所以 cost-reported 语义汇入
+     *  `judgeCostReportedByExecutor`，不再单独引一个 VariantResult flag。
      *  v0.30 新增,旧报告无此字段 — 老 costUSD 仍等于 execCostUSD + judgeCostUSD,新 costUSD
      *  含 diagnostic,跨版本汇总时按字段是否存在判断。 */
     diagnosticCostUSD?: number;
@@ -32,12 +35,18 @@ export interface VariantResult {
      *  report cost (currently codex). Default undefined ⇒ reported. */
     judgeCostReportedByExecutor?: boolean;
     numTurns: number;
+    /** Executor attempts used to obtain the final output. Missing means one. */
+    attemptCount?: number;
     fullNumTurns?: number;
     numSubAgents?: number;
     assistantTurns?: number;
     toolTurns?: number;
     numToolCalls?: number;
     numToolFailures?: number;
+    /** Calls explicitly cancelled by the runtime; distinct from execution failures. */
+    numToolCancelled?: number;
+    /** Calls without an authoritative completion state; excluded from success-rate denominator. */
+    numToolUnknown?: number;
     toolSuccessRate?: number;
     toolNames?: string[];
     /** per-sample tool call distribution (tool name → call count).
@@ -93,6 +102,9 @@ export interface VariantResult {
     timing?: {
         execMs: number;
         gradeMs: number;
+        /** Wall-clock time spent on the optional failure diagnostic call. */
+        diagnosticMs?: number;
+        /** execMs + gradeMs + (diagnosticMs ?? 0). */
         totalMs: number;
     };
 }
@@ -105,6 +117,12 @@ export interface VariantSummary {
     avgInputTokens: number;
     avgOutputTokens: number;
     avgTotalTokens: number;
+    /**
+     * Share of all sample results whose executor reported token usage.
+     * Missing means legacy reports where usage was assumed fully observed.
+     * Token averages use only successful results with observed usage.
+     */
+    tokenUsageCoverageRate?: number;
     /** sum(ok-sample 的 costUSD)。等于 totalExecCostUSD + totalJudgeCostUSD + totalDiagnosticCostUSD。
      *  注意:仅含执行成功且未被 per-sample budget 标 overrun 的 sample,跟 meta.totalCostUSD(全量
      *  累计,含失败 sample)语义不同 — 那是历史 ok-filter 行为,跟 v0.30 的 diagnostic 引入无关。 */
@@ -132,6 +150,8 @@ export interface VariantSummary {
     avgToolTurns?: number;
     avgToolCalls?: number;
     avgToolFailures?: number;
+    avgToolCancelled?: number;
+    avgToolUnknown?: number;
     toolSuccessRate?: number;
     toolDistribution?: Record<string, number>;
     traceCoverageRate?: number;
@@ -264,14 +284,28 @@ export interface ReportMeta {
      *  Value 2 marked the artifactHashes tree-hash era for local dir-skills (git dir-skills still
      *  SKILL.md-only). Value 3 extends whole-tree hashing to git dir-skills via isolated copies
      *  (all dir-skills bind). Value 4 keeps the v3 hash/binding semantics but marks the canonical
-     *  top-level discriminant era, so external consumers can version-gate the JSON shape. Drift /
-     *  lineage consumers gate on `>= 2` (tree-hash era); git-dir-skill binding additionally
-     *  requires `>= 3`. */
+     *  top-level discriminant era, so external consumers can version-gate the JSON shape. Value 5
+     *  changes `sampleHashes` to bind the complete Sample contract plus external mock fixtures and
+     *  the statically resolvable ESM custom-assertion import graph. Drift / lineage consumers gate on `>= 2`
+     *  (tree-hash era); git-dir-skill binding additionally requires `>= 3`; complete sample
+     *  fingerprints require `>= 5`. */
     schemaVersion?: number;
-    /** SHA256-12 of every sample's content (sample_id → hash). Same hash = same sample. */
+    /** SHA256-12 of every sample's measurement contract (sample_id → hash).
+     *  schemaVersion >= 5 binds all inline fields, mock return_file content, and the statically
+     *  resolvable ESM custom-assertion import graph. Earlier versions used a partial inline-field
+     *  hash. */
     sampleHashes?: Record<string, string>;
     /** SHA256-12 of the LLM judge prompt template. Different hash = judge changed semantics. */
     judgePromptHash?: string;
+    /** Failure-diagnostic execution contract. Diagnostic is independent from judge scoring,
+     *  so it remains configured under --no-judge unless --no-diagnostic is set. */
+    diagnostic?: {
+        enabled: boolean;
+        executor?: string;
+        model?: string;
+        runtime?: ExecutorRuntimeFingerprint;
+        promptHash?: string;
+    };
     /** Runtime fingerprint for the executor that produced tested outputs.
      *  Legacy/common field. Prefer executorRuntimes for variant-level audit; this
      *  field is the common runtime when all variants match, otherwise a representative
@@ -331,6 +365,17 @@ export interface ReportMeta {
     evolve?: {
         skillName: string;
         skillPath?: string;
+        /** 本次 evolve 实际发生的改写、修复与评测总成本；不同于报告内 result 的测量成本。 */
+        processCostUSD?: number;
+        /** false 表示 processCostUSD 只是可观测 provider 成本的下界。 */
+        processCostReported?: boolean;
+        /** 每个 round 的原始评测身份。合并报告是 aggregate，不复用任一源报告的 request / run / job。 */
+        sourceReports?: Array<{
+            round: number;
+            accepted: boolean;
+            reportId: string;
+            variant: string;
+        }>;
     };
 }
 export interface ResultEntry {
@@ -354,10 +399,17 @@ export interface ResultEntry {
 export interface SampleSnapshot {
     sample_id: string;
     prompt: string;
+    /** Runtime context that can change executor-visible project state. */
+    cwd?: string;
     rubric?: string;
     context?: string;
+    dimensions?: Record<string, string>;
     assertions?: import('./eval.js').Assertion[];
     mocks?: import('./eval.js').Mock[];
+    mocksStrict?: boolean;
+    environment?: import('./eval.js').SampleEnvironment;
+    allowedTools?: string[];
+    expectedTools?: string[];
     capability?: string[];
     difficulty?: import('./eval.js').SampleDifficulty;
     construct?: string;

package/dist/types/skill-index.d.ts

@@ -36,8 +36,13 @@ export interface SkillObserveSnapshot {
     generatedAt: string;
     healthBand: 'green' | 'yellow' | 'red';
     failureRate: number;
+    toolCallCount?: number;
+    toolResolvedCount?: number;
+    toolCancelledCount?: number;
+    toolUnknownCount?: number;
     segmentCount: number;
     gapRate: number;
+    stability?: 'stable' | 'unstable' | 'very-unstable' | 'unknown';
     /** 统计可信度(按 segment 数)。underpowered 时下游 insight / card 不应触发硬红或 high severity。
      *  历史快照缺此字段时由 segmentCount 兜底推导。 */
     confidence: 'high' | 'low' | 'underpowered';
@@ -56,6 +61,8 @@ export interface SkillGraphNodePreview {
     nodeKind: string;
     label: string;
     status?: string;
+    /** Studio evidence view 用于把 assertion / diagnostic 归到对应 sample。 */
+    parentSampleStableKey?: string;
     coverage?: 'declared' | 'undeclared';
     coveredBySamples?: string[];
 }

package/dist/types/trace.d.ts

@@ -0,0 +1,2 @@
+/** Stable source identity shared by execution evidence and Trace IR. */
+export type TraceSourceKind = 'claude' | 'codex' | 'openclaw' | 'markdown_log' | 'unknown';

package/dist/types/trace.js

@@ -0,0 +1 @@
+export {};

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "oh-my-knowledge",
-  "version": "0.48.0",
+  "version": "0.49.0",
   "packageManager": "yarn@4.16.0",
   "description": "Evaluation framework for LLM knowledge inputs — prompts, RAG corpora, skills, agent workflows. Fix the model, vary the artifact. Built-in statistical rigor: bootstrap CI, Krippendorff α, length-debias, saturation curves.",
   "type": "module",
@@ -76,6 +76,9 @@
     "skill-evaluation",
     "knowledge-engineering",
     "prompt-engineering",
+    "codex",
+    "codex-cli",
+    "openai",
     "claude-code",
     "claude",
     "llm",
@@ -93,13 +96,14 @@
   "license": "MIT",
   "dependencies": {
     "@anthropic-ai/claude-agent-sdk": "^0.3.143",
-    "@anthropic-ai/sdk": "^0.107.0",
+    "@anthropic-ai/sdk": "^0.112.3",
     "@inquirer/prompts": "^8.4.3",
     "@modelcontextprotocol/sdk": "^1.29.0",
     "@oclif/core": "^4",
-    "@openai/codex-sdk": "0.142.4",
+    "@openai/codex-sdk": "0.144.6",
     "ajv": "^8.18.0",
     "chart.js": "^4.5.1",
+    "es-module-lexer": "^2.0.0",
     "js-yaml": "^4.1.1",
     "simple-statistics": "^7.8.9",
     "zod": "^4.4.3"
@@ -112,11 +116,11 @@
     "@types/node": "^25.5.0",
     "eslint": "^10.1.0",
     "husky": "^9.1.7",
-    "lint-staged": "17.0.8",
+    "lint-staged": "17.1.0",
     "npm-run-all2": "^9.0.1",
     "typescript": "^6.0.2",
     "typescript-eslint": "^8.58.0",
     "vitepress": "^1.6.4",
-    "vitest": "4.1.9"
+    "vitest": "4.1.10"
   }
 }