openclaw-cortex-memory 0.1.0-Alpha.4 → 0.1.0-Alpha.5

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (43) hide show
  1. package/README.md +82 -3
  2. package/SKILL.md +58 -3
  3. package/dist/index.d.ts +18 -0
  4. package/dist/index.d.ts.map +1 -1
  5. package/dist/index.js +118 -2
  6. package/dist/index.js.map +1 -1
  7. package/dist/openclaw.plugin.json +101 -1
  8. package/dist/src/dedup/three_stage_deduplicator.d.ts +25 -0
  9. package/dist/src/dedup/three_stage_deduplicator.d.ts.map +1 -0
  10. package/dist/src/dedup/three_stage_deduplicator.js +225 -0
  11. package/dist/src/dedup/three_stage_deduplicator.js.map +1 -0
  12. package/dist/src/engine/ts_engine.d.ts +36 -0
  13. package/dist/src/engine/ts_engine.d.ts.map +1 -1
  14. package/dist/src/engine/ts_engine.js +197 -32
  15. package/dist/src/engine/ts_engine.js.map +1 -1
  16. package/dist/src/engine/types.d.ts +4 -0
  17. package/dist/src/engine/types.d.ts.map +1 -1
  18. package/dist/src/graph/ontology.d.ts +53 -0
  19. package/dist/src/graph/ontology.d.ts.map +1 -0
  20. package/dist/src/graph/ontology.js +252 -0
  21. package/dist/src/graph/ontology.js.map +1 -0
  22. package/dist/src/session/session_end.d.ts +55 -0
  23. package/dist/src/session/session_end.d.ts.map +1 -1
  24. package/dist/src/session/session_end.js +237 -51
  25. package/dist/src/session/session_end.js.map +1 -1
  26. package/dist/src/store/archive_store.d.ts +89 -0
  27. package/dist/src/store/archive_store.d.ts.map +1 -0
  28. package/dist/src/store/archive_store.js +242 -0
  29. package/dist/src/store/archive_store.js.map +1 -0
  30. package/dist/src/store/read_store.d.ts +24 -0
  31. package/dist/src/store/read_store.d.ts.map +1 -1
  32. package/dist/src/store/read_store.js +635 -27
  33. package/dist/src/store/read_store.js.map +1 -1
  34. package/dist/src/store/vector_store.d.ts +30 -0
  35. package/dist/src/store/vector_store.d.ts.map +1 -0
  36. package/dist/src/store/vector_store.js +127 -0
  37. package/dist/src/store/vector_store.js.map +1 -0
  38. package/dist/src/sync/session_sync.d.ts +7 -0
  39. package/dist/src/sync/session_sync.d.ts.map +1 -1
  40. package/dist/src/sync/session_sync.js +109 -7
  41. package/dist/src/sync/session_sync.js.map +1 -1
  42. package/openclaw.plugin.json +101 -1
  43. package/package.json +6 -2
@@ -114,12 +114,35 @@ function parseJsonlFile(filePath, sourceLabel, logger) {
114
114
  }
115
115
  const id = typeof parsed.id === "string" ? parsed.id : `${sourceLabel}:${docs.length + 1}`;
116
116
  const timestampValue = typeof parsed.timestamp === "string" ? Date.parse(parsed.timestamp) : NaN;
117
+ const entities = Array.isArray(parsed.entities)
118
+ ? parsed.entities.map(item => (typeof item === "string" ? item.trim() : "")).filter(Boolean)
119
+ : [];
120
+ const relations = Array.isArray(parsed.relations)
121
+ ? parsed.relations
122
+ .map(item => {
123
+ if (typeof item !== "object" || item === null)
124
+ return null;
125
+ const relation = item;
126
+ const source = typeof relation.source === "string" ? relation.source.trim() : "";
127
+ const target = typeof relation.target === "string" ? relation.target.trim() : "";
128
+ const type = typeof relation.type === "string" ? relation.type.trim() : "related_to";
129
+ if (!source || !target)
130
+ return null;
131
+ return { source, target, type };
132
+ })
133
+ .filter((item) => Boolean(item))
134
+ : [];
117
135
  docs.push({
118
136
  id,
119
137
  text,
120
138
  source: sourceLabel,
121
139
  timestamp: Number.isFinite(timestampValue) ? timestampValue : undefined,
122
140
  embedding: Array.isArray(parsed.embedding) ? parsed.embedding.filter(item => Number.isFinite(item)) : undefined,
141
+ eventType: typeof parsed.event_type === "string" ? parsed.event_type.trim() : undefined,
142
+ qualityScore: typeof parsed.quality_score === "number" ? parsed.quality_score : undefined,
143
+ sessionId: typeof parsed.session_id === "string" ? parsed.session_id : undefined,
144
+ entities,
145
+ relations,
123
146
  });
124
147
  }
125
148
  catch (error) {
@@ -161,6 +184,78 @@ function withRecencyBoost(score, timestamp) {
161
184
  }
162
185
  return score;
163
186
  }
187
+ function recencyScore(timestamp) {
188
+ if (!timestamp) {
189
+ return 0;
190
+ }
191
+ const ageHours = (Date.now() - timestamp) / (1000 * 60 * 60);
192
+ if (ageHours < 12)
193
+ return 1;
194
+ if (ageHours < 24)
195
+ return 0.8;
196
+ if (ageHours < 72)
197
+ return 0.6;
198
+ if (ageHours < 168)
199
+ return 0.4;
200
+ if (ageHours < 720)
201
+ return 0.2;
202
+ return 0.05;
203
+ }
204
+ function eventTypeHalfLifeDays(eventType, options) {
205
+ const fallback = typeof options?.defaultHalfLifeDays === "number" && options.defaultHalfLifeDays > 0
206
+ ? options.defaultHalfLifeDays
207
+ : 90;
208
+ const type = (eventType || "").trim().toLowerCase();
209
+ if (!type)
210
+ return fallback;
211
+ const configured = options?.halfLifeByEventType || {};
212
+ if (typeof configured[type] === "number" && configured[type] > 0) {
213
+ return configured[type];
214
+ }
215
+ if (["issue", "fix", "action_item", "blocker"].includes(type))
216
+ return 30;
217
+ if (["plan", "milestone", "follow_up"].includes(type))
218
+ return 60;
219
+ if (["decision", "insight", "retrospective"].includes(type))
220
+ return 120;
221
+ if (["preference", "constraint", "requirement", "dependency", "assumption"].includes(type))
222
+ return 240;
223
+ return fallback;
224
+ }
225
+ function computeAntiDecayBoost(id, hitStats, options) {
226
+ const anti = options?.antiDecay;
227
+ if (anti?.enabled === false) {
228
+ return 1;
229
+ }
230
+ const item = hitStats.items[id];
231
+ if (!item) {
232
+ return 1;
233
+ }
234
+ const hitWeight = typeof anti?.hitWeight === "number" && anti.hitWeight > 0 ? anti.hitWeight : 0.08;
235
+ const maxBoost = typeof anti?.maxBoost === "number" && anti.maxBoost >= 1 ? anti.maxBoost : 1.6;
236
+ const recentWindowDays = typeof anti?.recentWindowDays === "number" && anti.recentWindowDays > 0 ? anti.recentWindowDays : 30;
237
+ const lastHitTs = Date.parse(item.lastHitAt || "");
238
+ const ageDays = Number.isFinite(lastHitTs) ? Math.max(0, (Date.now() - lastHitTs) / (1000 * 60 * 60 * 24)) : recentWindowDays * 2;
239
+ const freshness = ageDays <= recentWindowDays ? (1 - ageDays / recentWindowDays) : 0;
240
+ const countFactor = Math.log1p(Math.max(0, item.count));
241
+ const boost = 1 + countFactor * hitWeight * (0.5 + 0.5 * freshness);
242
+ return Math.min(maxBoost, Math.max(1, boost));
243
+ }
244
+ function computeDecayFactor(id, eventType, timestamp, options, hitStats) {
245
+ const enabled = options?.enabled !== false;
246
+ if (!enabled || !timestamp) {
247
+ return computeAntiDecayBoost(id, hitStats, options);
248
+ }
249
+ const ageDays = Math.max(0, (Date.now() - timestamp) / (1000 * 60 * 60 * 24));
250
+ const halfLife = eventTypeHalfLifeDays(eventType, options);
251
+ const base = Math.pow(2, -ageDays / Math.max(1, halfLife));
252
+ const floor = typeof options?.minFloor === "number"
253
+ ? Math.max(0, Math.min(1, options.minFloor))
254
+ : 0.15;
255
+ const decay = Math.max(floor, base);
256
+ const boost = computeAntiDecayBoost(id, hitStats, options);
257
+ return Math.min(1, decay * boost);
258
+ }
164
259
  function normalizeBaseUrl(value) {
165
260
  if (!value)
166
261
  return "";
@@ -287,8 +382,314 @@ async function requestRerank(args) {
287
382
  }
288
383
  throw lastError instanceof Error ? lastError : new Error(String(lastError || "rerank_failed"));
289
384
  }
385
+ function classifyIntent(query) {
386
+ const text = query.toLowerCase();
387
+ const relationHints = /(关系|依赖|关联|上下游|graph|relation|entity|拓扑)/i;
388
+ if (relationHints.test(text))
389
+ return "RELATION_DISCOVERY";
390
+ const troubleHints = /(报错|错误|异常|失败|超时|无法|崩溃|error|failed|timeout|fix)/i;
391
+ if (troubleHints.test(text))
392
+ return "TROUBLESHOOTING";
393
+ const preferenceHints = /(偏好|习惯|口味|喜欢|不喜欢|偏向|preference)/i;
394
+ if (preferenceHints.test(text))
395
+ return "PREFERENCE_PROFILE";
396
+ const timelineHints = /(最近|上次|之前|时间线|timeline|history)/i;
397
+ if (timelineHints.test(text))
398
+ return "TIMELINE_REVIEW";
399
+ const decisionHints = /(方案|决策|选择|建议|取舍|tradeoff|plan)/i;
400
+ if (decisionHints.test(text))
401
+ return "DECISION_SUPPORT";
402
+ return "FACT_LOOKUP";
403
+ }
404
+ function preferredEventTypes(intent) {
405
+ if (intent === "TROUBLESHOOTING")
406
+ return ["issue", "fix", "risk", "blocker", "dependency", "retrospective"];
407
+ if (intent === "PREFERENCE_PROFILE")
408
+ return ["preference", "decision", "constraint", "requirement"];
409
+ if (intent === "DECISION_SUPPORT")
410
+ return ["decision", "plan", "insight", "assumption", "constraint", "requirement"];
411
+ if (intent === "TIMELINE_REVIEW")
412
+ return ["action_item", "follow_up", "milestone", "plan", "decision", "issue", "fix"];
413
+ return [];
414
+ }
415
+ function sourceWeight(source, intent) {
416
+ if (source === "rules") {
417
+ return intent === "DECISION_SUPPORT" || intent === "TROUBLESHOOTING" ? 1.15 : 0.9;
418
+ }
419
+ if (source === "graph") {
420
+ return intent === "RELATION_DISCOVERY" ? 1.25 : 0.85;
421
+ }
422
+ if (source === "vector") {
423
+ return 1.05;
424
+ }
425
+ return 1;
426
+ }
427
+ async function searchLanceDb(args) {
428
+ try {
429
+ const lancedbDir = path.join(args.memoryRoot, "vector", "lancedb");
430
+ if (!fs.existsSync(lancedbDir)) {
431
+ return [];
432
+ }
433
+ const dynamicImport = new Function("specifier", "return import(specifier)");
434
+ const moduleValue = await dynamicImport("@lancedb/lancedb");
435
+ const connect = moduleValue.connect;
436
+ if (typeof connect !== "function") {
437
+ return [];
438
+ }
439
+ const db = await connect(lancedbDir);
440
+ if (!db || typeof db.openTable !== "function") {
441
+ return [];
442
+ }
443
+ const table = await db.openTable("events");
444
+ if (!table || typeof table.search !== "function") {
445
+ return [];
446
+ }
447
+ const searchObj = table.search(args.queryEmbedding);
448
+ if (!searchObj || typeof searchObj.limit !== "function") {
449
+ return [];
450
+ }
451
+ const limited = searchObj.limit(args.limit);
452
+ if (!limited || typeof limited.toArray !== "function") {
453
+ return [];
454
+ }
455
+ const rows = await limited.toArray();
456
+ const docs = [];
457
+ for (const row of rows) {
458
+ if (typeof row !== "object" || row === null)
459
+ continue;
460
+ const record = row;
461
+ const id = typeof record.id === "string" ? record.id : "";
462
+ const summary = typeof record.summary === "string" ? record.summary : "";
463
+ if (!id || !summary)
464
+ continue;
465
+ const ts = typeof record.timestamp === "string" ? Date.parse(record.timestamp) : NaN;
466
+ const entities = typeof record.entities_json === "string"
467
+ ? JSON.parse(record.entities_json).filter(item => typeof item === "string" && item.trim())
468
+ : [];
469
+ const relations = typeof record.relations_json === "string"
470
+ ? JSON.parse(record.relations_json)
471
+ : [];
472
+ docs.push({
473
+ id,
474
+ text: summary,
475
+ source: "vector_lancedb",
476
+ timestamp: Number.isFinite(ts) ? ts : undefined,
477
+ embedding: Array.isArray(record.vector) ? record.vector.filter(item => Number.isFinite(item)) : undefined,
478
+ eventType: typeof record.event_type === "string" ? record.event_type : undefined,
479
+ qualityScore: typeof record.quality_score === "number" ? record.quality_score : undefined,
480
+ sessionId: typeof record.session_id === "string" ? record.session_id : undefined,
481
+ entities,
482
+ relations: Array.isArray(relations) ? relations : [],
483
+ });
484
+ }
485
+ return docs;
486
+ }
487
+ catch (error) {
488
+ args.logger.debug(`LanceDB search fallback: ${error}`);
489
+ return [];
490
+ }
491
+ }
492
+ function parseVectorFallback(filePath, logger) {
493
+ const content = safeReadFile(filePath);
494
+ if (!content) {
495
+ return [];
496
+ }
497
+ const docs = [];
498
+ for (const line of content.split(/\r?\n/)) {
499
+ const trimmed = line.trim();
500
+ if (!trimmed)
501
+ continue;
502
+ try {
503
+ const parsed = JSON.parse(trimmed);
504
+ const id = typeof parsed.id === "string" ? parsed.id : "";
505
+ const summary = typeof parsed.summary === "string" ? parsed.summary.trim() : "";
506
+ if (!id || !summary)
507
+ continue;
508
+ const ts = typeof parsed.timestamp === "string" ? Date.parse(parsed.timestamp) : NaN;
509
+ const entities = Array.isArray(parsed.entities)
510
+ ? parsed.entities.map(item => (typeof item === "string" ? item.trim() : "")).filter(Boolean)
511
+ : [];
512
+ const relations = Array.isArray(parsed.relations)
513
+ ? parsed.relations
514
+ .map(item => {
515
+ if (typeof item !== "object" || item === null)
516
+ return null;
517
+ const relation = item;
518
+ const source = typeof relation.source === "string" ? relation.source.trim() : "";
519
+ const target = typeof relation.target === "string" ? relation.target.trim() : "";
520
+ const type = typeof relation.type === "string" ? relation.type.trim() : "related_to";
521
+ if (!source || !target)
522
+ return null;
523
+ return { source, target, type };
524
+ })
525
+ .filter((item) => Boolean(item))
526
+ : [];
527
+ docs.push({
528
+ id,
529
+ text: summary,
530
+ source: "vector_jsonl",
531
+ timestamp: Number.isFinite(ts) ? ts : undefined,
532
+ embedding: Array.isArray(parsed.embedding) ? parsed.embedding.filter(item => Number.isFinite(item)) : undefined,
533
+ eventType: typeof parsed.event_type === "string" ? parsed.event_type.trim() : undefined,
534
+ qualityScore: typeof parsed.quality_score === "number" ? parsed.quality_score : undefined,
535
+ sessionId: typeof parsed.session_id === "string" ? parsed.session_id : undefined,
536
+ entities,
537
+ relations,
538
+ });
539
+ }
540
+ catch (error) {
541
+ logger.debug(`Skip invalid vector jsonl line: ${error}`);
542
+ }
543
+ }
544
+ return docs;
545
+ }
546
+ async function requestFusion(args) {
547
+ const endpoint = args.llm.baseUrl.endsWith("/chat/completions")
548
+ ? args.llm.baseUrl
549
+ : `${args.llm.baseUrl}/chat/completions`;
550
+ const evidenceText = args.candidates
551
+ .map((item, index) => `${index + 1}. [${item.id}] (${item.source}, score=${item.score.toFixed(4)}) ${item.text}`)
552
+ .join("\n")
553
+ .slice(0, 18000);
554
+ const prompt = [
555
+ "你是记忆检索融合器。请融合多路召回结果,产出可直接给 Agent 使用的完整记忆包,不要让 Agent 再去翻历史。",
556
+ "必须严格返回 JSON:",
557
+ "{\"canonical_answer\": string, \"coverage_note\": string, \"facts\": [{\"text\": string, \"evidence_ids\": string[]}], \"timeline\": [{\"when\": string, \"event\": string, \"evidence_ids\": string[]}], \"entities\": [{\"name\": string, \"role\": string}], \"decisions\": [{\"decision\": string, \"rationale\": string, \"evidence_ids\": string[]}], \"fixes\": [{\"issue\": string, \"fix\": string, \"evidence_ids\": string[]}], \"preferences\": [{\"subject\": string, \"preference\": string, \"evidence_ids\": string[]}], \"risks\": [{\"risk\": string, \"mitigation\": string, \"evidence_ids\": string[]}], \"action_items\": [{\"item\": string, \"owner\": string, \"status\": string, \"evidence_ids\": string[]}], \"conflicts\": [{\"topic\": string, \"details\": string}], \"evidence_ids\": string[], \"confidence\": number}",
558
+ "要求:",
559
+ "1) canonical_answer 是完整可执行答案,不要只写摘要",
560
+ "2) facts 3-12 条,优先高分证据",
561
+ "3) evidence_ids 必须来自输入候选 id",
562
+ "4) 若存在冲突写入 conflicts,否则返回空数组",
563
+ "5) confidence 0~1",
564
+ "6) 不确定信息必须在 coverage_note 标注",
565
+ ].join("\n");
566
+ const body = {
567
+ model: args.llm.model,
568
+ temperature: 0.1,
569
+ messages: [
570
+ { role: "system", content: "你只输出 JSON,不要额外解释。" },
571
+ { role: "user", content: `${prompt}\n\n问题:\n${args.query}\n\n候选证据:\n${evidenceText}` },
572
+ ],
573
+ };
574
+ let lastError = null;
575
+ for (let attempt = 0; attempt < 2; attempt += 1) {
576
+ const controller = new AbortController();
577
+ const timeoutId = setTimeout(() => controller.abort(), 20000);
578
+ try {
579
+ const response = await fetch(endpoint, {
580
+ method: "POST",
581
+ headers: {
582
+ "content-type": "application/json",
583
+ authorization: `Bearer ${args.llm.apiKey}`,
584
+ },
585
+ body: JSON.stringify(body),
586
+ signal: controller.signal,
587
+ });
588
+ clearTimeout(timeoutId);
589
+ if (!response.ok) {
590
+ lastError = new Error(`fusion_http_${response.status}`);
591
+ continue;
592
+ }
593
+ const json = await response.json();
594
+ const content = json?.choices?.[0]?.message?.content?.trim() || "";
595
+ if (!content) {
596
+ lastError = new Error("fusion_empty");
597
+ continue;
598
+ }
599
+ const parsed = JSON.parse(content);
600
+ if (!parsed || typeof parsed.canonical_answer !== "string" || !parsed.canonical_answer.trim()) {
601
+ lastError = new Error("fusion_invalid");
602
+ continue;
603
+ }
604
+ const evidenceIds = Array.isArray(parsed.evidence_ids)
605
+ ? parsed.evidence_ids.filter(item => typeof item === "string" && item.trim())
606
+ : [];
607
+ return {
608
+ canonical_answer: parsed.canonical_answer.trim().slice(0, 6000),
609
+ coverage_note: typeof parsed.coverage_note === "string" ? parsed.coverage_note.trim().slice(0, 1200) : "",
610
+ facts: Array.isArray(parsed.facts) ? parsed.facts : [],
611
+ timeline: Array.isArray(parsed.timeline) ? parsed.timeline : [],
612
+ entities: Array.isArray(parsed.entities) ? parsed.entities : [],
613
+ decisions: Array.isArray(parsed.decisions) ? parsed.decisions : [],
614
+ fixes: Array.isArray(parsed.fixes) ? parsed.fixes : [],
615
+ preferences: Array.isArray(parsed.preferences) ? parsed.preferences : [],
616
+ risks: Array.isArray(parsed.risks) ? parsed.risks : [],
617
+ action_items: Array.isArray(parsed.action_items) ? parsed.action_items : [],
618
+ conflicts: Array.isArray(parsed.conflicts) ? parsed.conflicts : [],
619
+ evidence_ids: evidenceIds,
620
+ confidence: typeof parsed.confidence === "number"
621
+ ? Math.max(0, Math.min(1, parsed.confidence))
622
+ : 0.5,
623
+ };
624
+ }
625
+ catch (error) {
626
+ clearTimeout(timeoutId);
627
+ lastError = error;
628
+ }
629
+ }
630
+ throw lastError instanceof Error ? lastError : new Error(String(lastError || "fusion_failed"));
631
+ }
290
632
  function createReadStore(options) {
291
633
  const memoryRoot = options.dbPath ? path.resolve(options.dbPath) : path.join(options.projectRoot, "data", "memory");
634
+ const vectorFallbackPath = path.join(memoryRoot, "vector", "lancedb_events.jsonl");
635
+ const hitStatsPath = path.join(memoryRoot, ".read_hit_stats.json");
636
+ function loadHitStats() {
637
+ try {
638
+ if (!fs.existsSync(hitStatsPath)) {
639
+ return { items: {} };
640
+ }
641
+ const content = fs.readFileSync(hitStatsPath, "utf-8").trim();
642
+ if (!content) {
643
+ return { items: {} };
644
+ }
645
+ const parsed = JSON.parse(content);
646
+ if (!parsed || typeof parsed !== "object" || !parsed.items || typeof parsed.items !== "object") {
647
+ return { items: {} };
648
+ }
649
+ return parsed;
650
+ }
651
+ catch {
652
+ return { items: {} };
653
+ }
654
+ }
655
+ function saveHitStats(state) {
656
+ try {
657
+ const dir = path.dirname(hitStatsPath);
658
+ if (!fs.existsSync(dir)) {
659
+ fs.mkdirSync(dir, { recursive: true });
660
+ }
661
+ fs.writeFileSync(hitStatsPath, JSON.stringify(state, null, 2), "utf-8");
662
+ }
663
+ catch (error) {
664
+ options.logger.warn(`Failed to persist read hit stats: ${error}`);
665
+ }
666
+ }
667
+ function markHit(ids) {
668
+ if (!ids.length) {
669
+ return;
670
+ }
671
+ const state = loadHitStats();
672
+ const now = new Date().toISOString();
673
+ for (const id of ids) {
674
+ const key = (id || "").trim();
675
+ if (!key)
676
+ continue;
677
+ const prev = state.items[key];
678
+ state.items[key] = {
679
+ count: (prev?.count || 0) + 1,
680
+ lastHitAt: now,
681
+ };
682
+ }
683
+ const entries = Object.entries(state.items)
684
+ .sort((a, b) => {
685
+ const ta = Date.parse(a[1].lastHitAt || "");
686
+ const tb = Date.parse(b[1].lastHitAt || "");
687
+ return (Number.isFinite(tb) ? tb : 0) - (Number.isFinite(ta) ? ta : 0);
688
+ })
689
+ .slice(0, 20000);
690
+ state.items = Object.fromEntries(entries);
691
+ saveHitStats(state);
692
+ }
292
693
  function loadAllDocuments() {
293
694
  const cortexRulesPath = path.join(memoryRoot, "CORTEX_RULES.md");
294
695
  const memoryMdPath = path.join(memoryRoot, "MEMORY.md");
@@ -307,6 +708,9 @@ function createReadStore(options) {
307
708
  return { results: [] };
308
709
  }
309
710
  const docs = loadAllDocuments();
711
+ const hitStats = loadHitStats();
712
+ const intent = classifyIntent(query);
713
+ const preferredTypes = preferredEventTypes(intent);
310
714
  let queryEmbedding = null;
311
715
  const embeddingModel = options.embedding?.model || "";
312
716
  const embeddingApiKey = options.embedding?.apiKey || "";
@@ -325,34 +729,148 @@ function createReadStore(options) {
325
729
  options.logger.warn(`Embedding query failed, fallback to lexical search: ${error}`);
326
730
  }
327
731
  }
328
- const lexicalRanked = docs
732
+ const vectorDocsFromLance = queryEmbedding && queryEmbedding.length > 0
733
+ ? await searchLanceDb({ memoryRoot, queryEmbedding, limit: Math.max(20, args.topK * 8), logger: options.logger })
734
+ : [];
735
+ const vectorDocsFallback = vectorDocsFromLance.length > 0
736
+ ? []
737
+ : parseVectorFallback(vectorFallbackPath, options.logger);
738
+ const vectorDocs = [...vectorDocsFromLance, ...vectorDocsFallback];
739
+ const graphDocs = docs
740
+ .filter(doc => Array.isArray(doc.relations) && doc.relations.length > 0)
329
741
  .map(doc => {
330
- const lexicalScore = scoreText(query, doc.text);
331
- const semanticScore = queryEmbedding && Array.isArray(doc.embedding) && doc.embedding.length > 0
742
+ const graphText = [
743
+ doc.text,
744
+ ...(doc.relations || []).map(relation => `${relation.source} ${relation.type} ${relation.target}`),
745
+ ].join(" | ");
746
+ return {
747
+ ...doc,
748
+ text: graphText,
749
+ };
750
+ });
751
+ const rulesDocs = docs.filter(doc => doc.source === "CORTEX_RULES.md");
752
+ const archiveDocs = docs.filter(doc => doc.source.startsWith("sessions_"));
753
+ const combinedCandidates = [];
754
+ const channels = {
755
+ rules: [],
756
+ archive: [],
757
+ vector: [],
758
+ graph: [],
759
+ };
760
+ const evaluateDoc = (doc, source) => {
761
+ const lexical = scoreText(query, doc.text);
762
+ const semantic = queryEmbedding && Array.isArray(doc.embedding) && doc.embedding.length > 0
332
763
  ? Math.max(0, cosineSimilarity(queryEmbedding, doc.embedding) * 5)
333
764
  : 0;
334
- const hybrid = lexicalScore + semanticScore;
335
- const total = withRecencyBoost(hybrid, doc.timestamp);
336
- return { doc, score: total };
337
- })
338
- .filter(item => item.score > 0)
765
+ if (lexical <= 0 && semantic <= 0) {
766
+ return null;
767
+ }
768
+ const recency = recencyScore(doc.timestamp);
769
+ const quality = typeof doc.qualityScore === "number" ? Math.max(0, Math.min(1, doc.qualityScore)) : 0.5;
770
+ const typeMatch = preferredTypes.length > 0 && doc.eventType
771
+ ? (preferredTypes.includes(doc.eventType) ? 1 : 0)
772
+ : 0.5;
773
+ const graphMatch = source === "graph" ? 1 : 0;
774
+ const baseWeighted = (0.2 * lexical +
775
+ 0.3 * semantic +
776
+ 0.1 * recency +
777
+ 0.15 * quality +
778
+ 0.15 * typeMatch +
779
+ 0.1 * graphMatch) * sourceWeight(source, intent);
780
+ const decayFactor = computeDecayFactor(doc.id, doc.eventType, doc.timestamp, options.memoryDecay, hitStats);
781
+ const weighted = baseWeighted * decayFactor;
782
+ return {
783
+ doc,
784
+ source,
785
+ lexical,
786
+ semantic,
787
+ recency,
788
+ quality,
789
+ typeMatch,
790
+ graphMatch,
791
+ decayFactor,
792
+ weighted,
793
+ };
794
+ };
795
+ for (const doc of rulesDocs) {
796
+ const candidate = evaluateDoc(doc, "rules");
797
+ if (candidate)
798
+ channels.rules.push(candidate);
799
+ }
800
+ for (const doc of archiveDocs) {
801
+ const candidate = evaluateDoc(doc, "archive");
802
+ if (candidate)
803
+ channels.archive.push(candidate);
804
+ }
805
+ for (const doc of vectorDocs) {
806
+ const candidate = evaluateDoc(doc, "vector");
807
+ if (candidate)
808
+ channels.vector.push(candidate);
809
+ }
810
+ for (const doc of graphDocs) {
811
+ const candidate = evaluateDoc(doc, "graph");
812
+ if (candidate)
813
+ channels.graph.push(candidate);
814
+ }
815
+ for (const key of Object.keys(channels)) {
816
+ channels[key].sort((a, b) => b.weighted - a.weighted);
817
+ combinedCandidates.push(...channels[key].slice(0, Math.max(20, args.topK * 5)));
818
+ }
819
+ const rrfMap = new Map();
820
+ const weightedMap = new Map();
821
+ const rrfK = 60;
822
+ for (const key of Object.keys(channels)) {
823
+ const list = channels[key];
824
+ for (let i = 0; i < list.length; i += 1) {
825
+ const candidate = list[i];
826
+ const rrf = 1 / (rrfK + i + 1);
827
+ rrfMap.set(candidate.doc.id, (rrfMap.get(candidate.doc.id) || 0) + rrf);
828
+ const current = weightedMap.get(candidate.doc.id);
829
+ if (!current || candidate.weighted > current.weighted) {
830
+ weightedMap.set(candidate.doc.id, candidate);
831
+ }
832
+ }
833
+ }
834
+ const preRanked = [...weightedMap.values()]
835
+ .map(candidate => ({
836
+ id: candidate.doc.id,
837
+ text: candidate.doc.text,
838
+ source: candidate.doc.source,
839
+ event_type: candidate.doc.eventType || "",
840
+ quality_score: candidate.quality,
841
+ timestamp: candidate.doc.timestamp ? new Date(candidate.doc.timestamp).toISOString() : "",
842
+ score: candidate.weighted + (rrfMap.get(candidate.doc.id) || 0) * 1.5,
843
+ reason_tags: [
844
+ `intent:${intent.toLowerCase()}`,
845
+ candidate.semantic > 0 ? "vector_hit" : "lexical_hit",
846
+ candidate.typeMatch >= 1 ? "event_type_match" : "event_type_weak",
847
+ candidate.recency >= 0.8 ? "recent" : "historical",
848
+ candidate.quality >= 0.7 ? "high_quality" : "normal_quality",
849
+ candidate.decayFactor < 1 ? `decay:${candidate.decayFactor.toFixed(3)}` : "decay:1.000",
850
+ `source:${candidate.source}`,
851
+ ],
852
+ }))
339
853
  .sort((a, b) => b.score - a.score)
340
- .slice(0, Math.max(1, Math.max(args.topK, 12)))
341
- .map(item => ({
342
- id: item.doc.id,
343
- text: item.doc.text,
344
- source: item.doc.source,
345
- score: Number(item.score.toFixed(4)),
346
- }));
854
+ .slice(0, Math.max(1, Math.max(args.topK, 20)));
855
+ const lexicalRanked = preRanked
856
+ .map(doc => {
857
+ const boost = withRecencyBoost(doc.score, doc.timestamp ? Date.parse(doc.timestamp) : undefined);
858
+ return { ...doc, score: Number(boost.toFixed(4)) };
859
+ });
347
860
  const rerankerModel = options.reranker?.model || "";
348
861
  const rerankerApiKey = options.reranker?.apiKey || "";
349
862
  const rerankerBaseUrl = normalizeBaseUrl(options.reranker?.baseURL || options.reranker?.baseUrl);
350
- let ranked = lexicalRanked;
863
+ let rerankedSimple = lexicalRanked.map(item => ({
864
+ id: item.id,
865
+ text: item.text,
866
+ source: item.source,
867
+ score: item.score,
868
+ }));
351
869
  if (rerankerModel && rerankerApiKey && rerankerBaseUrl && lexicalRanked.length > 1) {
352
870
  try {
353
- ranked = await requestRerank({
871
+ rerankedSimple = await requestRerank({
354
872
  query,
355
- candidates: lexicalRanked,
873
+ candidates: lexicalRanked.map(item => ({ id: item.id, text: item.text, source: item.source, score: item.score })),
356
874
  model: rerankerModel,
357
875
  apiKey: rerankerApiKey,
358
876
  baseUrl: rerankerBaseUrl,
@@ -362,27 +880,103 @@ function createReadStore(options) {
362
880
  options.logger.warn(`Reranker failed, keep hybrid ranking: ${error}`);
363
881
  }
364
882
  }
365
- ranked = ranked.slice(0, Math.max(1, args.topK)).map(item => ({
366
- id: item.id,
367
- text: item.text,
368
- source: item.source,
369
- score: Number(item.score.toFixed(4)),
370
- }));
883
+ const ranked = rerankedSimple.slice(0, Math.max(1, args.topK)).map(item => {
884
+ const hit = lexicalRanked.find(entry => entry.id === item.id);
885
+ return {
886
+ id: item.id,
887
+ text: item.text,
888
+ source: item.source,
889
+ event_type: hit?.event_type || "",
890
+ quality_score: hit?.quality_score ?? 0,
891
+ timestamp: hit?.timestamp || "",
892
+ score: Number(item.score.toFixed(4)),
893
+ reason_tags: Array.isArray(hit?.reason_tags) ? hit?.reason_tags : [],
894
+ };
895
+ });
896
+ const fusionEnabled = options.fusion?.enabled !== false;
897
+ const llmModel = options.llm?.model || "";
898
+ const llmApiKey = options.llm?.apiKey || "";
899
+ const llmBaseUrl = normalizeBaseUrl(options.llm?.baseURL || options.llm?.baseUrl);
900
+ if (fusionEnabled && llmModel && llmApiKey && llmBaseUrl && ranked.length > 1) {
901
+ try {
902
+ const maxCandidates = Math.max(4, Math.min(20, options.fusion?.maxCandidates ?? 10));
903
+ const fusion = await requestFusion({
904
+ query,
905
+ candidates: ranked.slice(0, maxCandidates).map(item => ({
906
+ id: item.id,
907
+ text: item.text,
908
+ source: item.source,
909
+ event_type: item.event_type,
910
+ quality_score: item.quality_score,
911
+ timestamp: item.timestamp,
912
+ score: item.score,
913
+ reason_tags: Array.isArray(item.reason_tags) ? item.reason_tags : [],
914
+ })),
915
+ llm: {
916
+ model: llmModel,
917
+ apiKey: llmApiKey,
918
+ baseUrl: llmBaseUrl,
919
+ },
920
+ });
921
+ if (fusion && fusion.canonical_answer) {
922
+ const fusedItem = {
923
+ id: `fusion_${Date.now().toString(36)}`,
924
+ text: fusion.canonical_answer,
925
+ source: "llm_fusion",
926
+ event_type: "fusion",
927
+ quality_score: Number(fusion.confidence.toFixed(4)),
928
+ timestamp: new Date().toISOString(),
929
+ score: Number((Math.max(...ranked.map(item => item.score)) + 1).toFixed(4)),
930
+ reason_tags: ["llm_fused_authoritative", `evidence:${fusion.evidence_ids.length}`],
931
+ fused_coverage_note: fusion.coverage_note || "",
932
+ fused_facts: fusion.facts,
933
+ fused_timeline: fusion.timeline || [],
934
+ fused_entities: fusion.entities || [],
935
+ fused_decisions: fusion.decisions || [],
936
+ fused_fixes: fusion.fixes || [],
937
+ fused_preferences: fusion.preferences || [],
938
+ fused_risks: fusion.risks || [],
939
+ fused_action_items: fusion.action_items || [],
940
+ fused_conflicts: fusion.conflicts,
941
+ fused_evidence_ids: fusion.evidence_ids,
942
+ };
943
+ const authoritative = options.fusion?.authoritative !== false;
944
+ if (authoritative) {
945
+ markHit(Array.isArray(fusion.evidence_ids) ? fusion.evidence_ids : []);
946
+ return { results: [fusedItem] };
947
+ }
948
+ const merged = [fusedItem, ...ranked];
949
+ markHit([
950
+ ...(Array.isArray(fusion.evidence_ids) ? fusion.evidence_ids : []),
951
+ ...ranked.map(item => item.id),
952
+ ]);
953
+ return { results: merged.slice(0, Math.max(1, args.topK)) };
954
+ }
955
+ }
956
+ catch (error) {
957
+ options.logger.warn(`LLM fusion failed, fallback to reranked results: ${error}`);
958
+ }
959
+ }
960
+ markHit(ranked.map(item => item.id));
371
961
  return { results: ranked };
372
962
  }
373
963
  async function getHotContext(args) {
374
964
  const limit = Math.max(1, args.limit);
375
965
  const docs = loadAllDocuments();
376
966
  const coreRules = docs.find(doc => doc.source === "CORTEX_RULES.md");
377
- const sessionDocs = docs
378
- .filter(doc => doc.source.startsWith("sessions_"))
967
+ const archiveDocs = docs
968
+ .filter(doc => doc.source === "sessions_archive")
379
969
  .sort((a, b) => (b.timestamp ?? 0) - (a.timestamp ?? 0))
380
970
  .slice(0, limit);
971
+ const issueFixPairs = docs
972
+ .filter(doc => doc.source === "sessions_archive" && (doc.eventType === "issue" || doc.eventType === "fix"))
973
+ .sort((a, b) => (b.timestamp ?? 0) - (a.timestamp ?? 0))
974
+ .slice(0, 2);
381
975
  const result = [];
382
976
  if (coreRules) {
383
977
  result.push({ id: coreRules.id, text: coreRules.text, source: coreRules.source });
384
978
  }
385
- for (const doc of sessionDocs) {
979
+ for (const doc of [...issueFixPairs, ...archiveDocs]) {
386
980
  result.push({ id: doc.id, text: doc.text, source: doc.source });
387
981
  }
388
982
  return { context: result.slice(0, limit) };
@@ -396,6 +990,20 @@ function createReadStore(options) {
396
990
  age_seconds: args.cachedAutoSearch.ageSeconds,
397
991
  };
398
992
  }
993
+ if (!result.auto_search) {
994
+ const docs = loadAllDocuments()
995
+ .filter(doc => doc.source === "sessions_archive" && doc.sessionId === args.sessionId)
996
+ .sort((a, b) => (b.timestamp ?? 0) - (a.timestamp ?? 0));
997
+ const latest = docs[0];
998
+ if (latest && latest.text.trim()) {
999
+ const light = await searchMemory({ query: latest.text.slice(0, 80), topK: 3 });
1000
+ result.auto_search = {
1001
+ query: latest.text.slice(0, 80),
1002
+ results: light.results,
1003
+ age_seconds: 0,
1004
+ };
1005
+ }
1006
+ }
399
1007
  if (args.includeHot) {
400
1008
  const hot = await getHotContext({ limit: 20 });
401
1009
  result.hot_context = hot.context;